Effects of Attack Value

[FactionHeir]

The 25 delay thing doesn't seem to do anything from my tests that I ran a few days back. Had DEK vs DFrK with DEK having a delay of 1 for both primary and secondary and the results were fairly even (well, the DFrK always won by a small margin actually)

Death anims. I don't really think so. With the spearman bash face/stab back anim, he only needs to start that animation and you can move him away and the enemy who was supposed to be hit/was just hit starts crawling on the floor anyway and completes his death anim without the actual jab in the back.

You still did not explain my findings above

[Kobal2fr]

The 25 delay thing doesn't seem to do anything from my tests that I ran a few days back. Had DEK vs DFrK with DEK having a delay of 1 for both primary and secondary and the results were fairly even (well, the DFrK always won by a small margin actually)

Yup, I think 25 is the fastest you can hope combat cycles to be, but I suppose one could set them to 50 to pace down combat, maybe ? So much to test...

Death anims. I don't really think so. With the spearman bash face/stab back anim, he only needs to start that animation and you can move him away and the enemy who was supposed to be hit/was just hit starts crawling on the floor anyway and completes his death anim without the actual jab in the back.

Well yes, because his death anim actually started with the bashed face. The engine tells the attacker to do anim A and the defender to do the corresponding death anim, and if anim A is interrupted somehow, the death anim will go on as ordered "in synch", only not anymore. It's not the same as anim A not happening at all.

You still did not explain my findings above

1. Note that if an AI unit routs on the walls, you can have the entire wall packed with fast swordsmen and hardly any of them will ever hit the routers who run through them.

From what I've witnessed, pathing on top of walls is frankly screwy. You can have a whole unit all bunched up as if in schiltron, every man trying to get some space so that the whole unit looks like it's pulsating, and have only one of the bunch actually fight. Or the opposite, men strung along two sets of walls and only two guys duking it out somewhere. Considering that the pathing for chasing routers is screwy enough on flat ground already...

Or, another possible explanation would be that there are bonuses to unit defense when they're defending their walls, and the swords actually get some hits in (without triggering anims, see answer to Didz above) but not enough to go through the upped defense. That may well be the case, because two similar units facing each other on top of a wall will usually result in the attackers getting trashed, and I don't think tower fire alone accounts for that (as towers shooting in melee kill a lot of defenders too). That's another test to set up : disable towers, give the AI one siege tower, defend, watch results.

Could be both too, of course.

2. When you walk a unit slowly in one direction and enemy routers run in the same direction against your unit, many of them die even though none of your units are facing them or even making any attempt at an attack animation. They just bump into the back of your unit and perish.

Like I said to Didz, I think units chasing routers are considered to be perma-charging, and charges don't always seem to trigger animations. There's mebbe some 2D collision detection going on in there as well, I think.

[FactionHeir]

For 1: The amount of units escaping does not semm too linked to unit type. DFK and Peasants escape at about the same rate on the walls.

For 2: Note my wording. The non routing unit (infantry) is not chasing the routers, it is walking in a direction only. The routers just bump into those guy's backs and die.

[Husar]

For 2: Note my wording. The non routing unit (infantry) is not chasing the routers, it is walking in a direction only. The routers just bump into those guy's backs and die.

I think that's because routers get the pansy trait which makes them give up as soon as they bump into an enemy soldier who is in pursuit mode. Try running your cavalry through some routers without being in pursuit mode, just run through, hardly anything will happen, I think that's what usually happens on the walls as well, no pursuit mode = noone gives up.

[Kobal2fr]

1.) Meh :/ That probably rules out defense factors then, but mebbe not screwy pathfinding.

2.) Then I suppose/conjecture there's an "attack of opportunity" system in place - units bumping into routers getting one or more "free strikes" against them, yet not triggering anims because it would disrupt the moving unit too much. If they all stopped and faced inward/sideways to strike at them, they'd present their backs to missiles/charges, need some time to reform properly etc..., making routers exploitable. Send a peasant unit in their back, make them withdraw through the enemy ranks, exploit the following chaos, sort of thing.
Remember back in STW/MTW when routing units utterly destroyed anything what got in the way of their mad dash to the edge of the map ? They probably learned from that mistake.

But "bumping" doesn't mean there's a real 3D collision model going on, a mere 2D circle around each individual soldiers is enough and easier to compute. TBH I'd be surprised if soldiers were more than 2D circles on the terrain for most if not all behind the scenes numbercrunching.

But... attacks of opportunity would conflict with point 1.) wouldn't they ? . I REALLY hate it when facts get in the way of perfectly sound theories .
Maybe said AoPs are disabled when units are too blobbed ? Or the routines forcing individual soldiers to get some personnal spacing conflict with/override them ? I also remember reading something about blobbed units getting huge stat penalties (mind you that was in early press releases and so on, might not have made it in for real)... Lots of conjectures, very little definite answers.

[gardibolt]

I am dumb enough. I once charged three units of hospitallers through a line of my own culverins just as they were firing - OUCH

I remember getting impatient with how long it was taking to blow down a tower with a bombard in a siege so I started heading a siege tower towards the wall.

Bad move. The siege tower moved faster than I thought/the bombard took longer than I thought to kill the tower and the siege tower (plus most of the men pushing it) got blown up instead.

[atheotes]

Remember back in STW/MTW when routing units utterly destroyed anything what got in the way of their mad dash to the edge of the map ? They probably learned from that mistake.

In RTW bridge battles sometimes units rout through my phalanx destroying their formation and leading to heavy casualties!!! Dont know if it happens in M2TW...Enemy units have not routed through mine in Bridge battles in M2TW.

On topic...Only if someone from CA would drop in and throw some light

[alpaca]

Cheers, Kobal. I concur
I ran a similar test to yours and it seems like the visuals are, well, visual

I pitted DFKized Highland Rabble against DFKs and their performance was quite similar (if you don't factor in stupid AI moves when I was using the DFKs, it ran back and forth) with an 8 vs 7 wins, most of them very close.

I second that there are probably special rules for charges where units don't have to play an animation.

FactionHeir: I can't really agree to your point about units not catching routers on walls, my observations were a bit different really. Maybe there's some effect there that keeps the soldiers from attacking.

As for the normal routers, I'll have to pay a bit closer attention to that.


However I want to throw another point into the discussion: Since soldiers are playing attack and death animations, and soldiers don't really die before finishing their death animation, there probably is an influence of the actual animation length to the unit's performance, and I think this could be what Palamedes meant when he once talked about good and bad animations in conjunction with unit performance. It's like "meh I just stabbed someone, gimme a break"
This could explain my experience that slow 2h axe units have a worse performance than sword or 1h axe units.

There actually is a kind of sphere around the units if I remember correctly what GrumpyOldMan and KnightErrant were talking about. This could be linked to personal space and I think it's linked to missile attack. A sphere is of course a much simpler collision mechanism than a normal 3d model.

Something else? Ah yes, blobbing: I still find it highly useful to blob my units all onto a point when defending or squeezing them through a gate or something. CA should have been much more rigid on this.
Some of my tests suggest that having two decent units in Schiltrom stand on top of each other is almost impregnable and can hold the line very well until your archers or javelinmen killed the enemy.

[Kobal2fr]

However I want to throw another point into the discussion: Since soldiers are playing attack and death animations, and soldiers don't really die before finishing their death animation, there probably is an influence of the actual animation length to the unit's performance, and I think this could be what Palamedes meant when he once talked about good and bad animations in conjunction with unit performance. It's like "meh I just stabbed someone, gimme a break"
This could explain my experience that slow 2h axe units have a worse performance than sword or 1h axe units.

Well, as I did a lot of billmen vs swords test I had to do a lot of idly staring at both movesets, and I'm not sure there's any clear speed advantage either way. Both movesets have fast attacks and slow attacks, protracted finishers and quick "bang, you're dead" ones in about equal measure. The swords do look faster on cursory glance because they seem to be able to cue a lot of slashes in a row but 1) they only do so when said slashes are parried/blocked ; and blocked/interrupted bill strikes also result in a flurry of bill moves sometimes and 2) swords seem to have more long finishers, when bill strike have a long preparation before the strike (arming their blow) but once the move is fired, it's very quick and the enemy keels over almost instantly.
Of course, short of obsessively spelunking in the animation files there's no clear way to know the proportion of fast to slow anims, their precise timing, or even if all anims are used in equal proportions, but from what I've seen they more or less even out. At least, when comparing sword+shield Vs 2h axe, but I'm not about to crosscheck every moveset. Even chronic anal retention has its limits .

The discrepancy Palamedes mentionned might have been this "quick attack, slow death" vs "long protracted attack, quick death" ? But that's not to say it's a discrepancy inter-units, could be intra-unit as well (this billman using anim A is more efficient that that billman using anim B). But if all anims are used in equal amounts, this wouldn't matter.

Has anyone witnessed attacks being actually interrupted ? I.e. one soldier is about to strike an enemy, but a second enemy cuts him down before he finishes ? If so, then that may be a point against longer attacks (I've never ever seen a soldier be attacked mid-finisher, nevermind killed)...but wether it's really significant is doubtfull in my book. But I'll setup another batch of these tests, this time using the vaunted über JHI anim, just to be sure.

In the meantime, I'd theorize that as of 1.2, 2H suck 1-on-1 because they have generally much lower defense (-4/6 for no shield, -3 for no sword parry bonus) and higher skeletal compensation (skimmed over the vanilla unit file, 2Hers are on the whole in the 1.25 - 1.35 range, all 1H axes/maces 1.2, all swords/pikes/halberds are 1) than 1Hers, coupled with the fact that their mass+charge bonus are usually not high enough to make up for them in sheer charge kills.

Wether that state of affairs is OK or not is another matter - after all, if they performed the same as their 1H+Shield counterparts there wouldn't be much point to them. Right now they're mainly tools suited to charge flanks/rear like so much cheap-and-slow cav, and to form a second line behind the main battleline, opportunistically charging in gaps/weak points. Using them as linemen is suicidal.
"Is that what we want them to be/what they historically were ?" is the real question.

[econ21]

Does anyone know what skeletal compensation is and why should it affect combat outcomes?

[FactionHeir]

Kobal: Try DEK vs NK with equalized everything. I have a feeling that billmen have a slightly better anim than the dismounted knights.

[Kobal2fr]

OK, did the test with JHI, and again the results fly in the face of my understanding. First, I pit them against Highland Rabble upgraded to JHI stats. Both units took similar losses on the charge, but the JHI always won the following brawl, with 20-30 men left. I figured "well, maybe they have a reach advantage. Also, the peasants are not designed to fight in highly_trained formation, that might be it".
The problem is that most western highly_trained units are pikemen, who wouldn't do. So next I pit Lamtuna Spearmen (with JHI stats) vs JHI. The fights were less lopsided, but still the JHI won quasi-all the time.

The thing is, all those fights followed the same math : units reduced to 45-50 on the charge, then both reduced to 30ish during the initial melee, then the JHI stop taking as many losses as their opponents. I looked even closer, and I noticed that sometimes, JHI "turn" attacks. That is to say, they do a normal attack, their enemy blocks, enemy starts to attack, but then the Janissary blocks it AND follows the block with a killing blow, in one fluid anim. Seems to happen more often when soldiers are spaced out than in close knit melees. Could be a mere "attack made to look like a counter", of course, but it really looks like it's triggered by the opponent attacking them.

DEK vs NS resulted in DEKs getting smashed all the time, wether the DEK had Sword stats or Swords having DEK stats.

So there *is* something about animations, and the initial result was a fluke. And I'm an idiot, but that went without saying.

However, I do believe it still proves that while attack is determined by anim+stats instead of stats alone, the defense stats have the same "value" no matter what animation set is used. Or, to put it another way, that models who have a shield are not inherently more resistant that models who don't - high defense 2 handers parry/block just as often as they do, only using the butt of their weapon instead of the shield. Which gives us a sound basis for balance : pitting units with the exact same defense would allow us the quantify how much "attack value" each anim set really has, and we could balance from there.
So my time was not TOTALLY wasted. Was it ?

[Husar]

I think the anims determine the attack frequency, higher frequency = more possible kills per minute = better overall performance.
You could take a stopwatch and take the time for every single attack animation of every unit type and then calculate the average attack animation time for that unit type, if they're the same, we have to find something else, if not, we have any answer.

[econ21]

So my time was not TOTALLY wasted. Was it ?

No, it's is good to know for sure, one way or the other.

Personally, I wish you had been right. It would be hard enough to balance the game on stats, without having to worry about quantifying the effects of animations.

And maybe I am alone on this, but I never see the animations. I guess I am jumping the view around and playing too zoomed out, but for me it is as if the animations don't exist. (I have the same problem with Dawn of War, which has great death animations, but is too frenetic to watch passively.)

[alpaca]

Kobal: Are you playing on medium? I have a feeling that on medium the player gets some advantage over the AI.

[Kobal2fr]

Well, in campaigns I always play on VH, but all of the tests were done in custom battles where there...is...no...

CURSE YOU, YOU HORRIBLE BEARDY GIT ! Now you've gone and made me notice that there *is* a difficulty setting in custom battles too, and that I left it on medium all the time, for all these tests. For all the battle tests I've *ever* done, in fact.

[Lusted]

Battle difficulty does not seem to affect custom battle testing, or at least from the tests i've run.

[alpaca]

Battle difficulty does not seem to affect custom battle testing, or at least from the tests i've run.

Well I don't know for sure, but it felt like I got slightly better results when playing on medium, it could just be coincidence or wishful thinking though

By the way, I'm not talking about the pre-campaign setting. There's a setting in the custom battle options. If that doesn't do anything why is it even there?

[FactionHeir]

As I posted elsewhere, the difficulty in the custom battles seem to affect the AI's overall morale, with VH having equal morale as the player.

[pike master]

just for the record very hard gives the ai an attack increase of plus five and a morale and stamina boost so i dont see how units can be balanced human versus ai with that setting.

[alpaca]

just for the record very hard gives the ai an attack increase of plus five and a morale and stamina boost so i dont see how units can be balanced human versus ai with that setting.

That's not true. It was like that in RTW but it isn't in M2TW

FactionHeir: I know that is how it supposedly works. But I just feel like the AI is always playing on vh.

[Alsn]

I realize that this might be the worst case of thread necromancy in the history of the internet and for that i apologize but I just had to get back to the discussion about routing units sometimes passing straight through enemies without suffering casualties.

I recently started a turks campaign and noted some very odd behavior when chasing down routers(HA + general armies so there was alot of routing going on).

When chasing with my general, if i targeted a routing unit and the general while on its way to that unit passed through another routing unit then the non-targeted unit would get captured.

However, when doing the same with turkomans or turkish horse archers, they would pass straight through the secondary unit.

Testing this further I loaded a scots save i had where i was getting sieged alot by the french(and as such had some looong trains of routers turned into chickens at my gates to chase down) and lo and behold, the same pattern emerged. Generals bodyguards would annihilate anything in its path while border horses would completely ignore any routers other than their assigned target.

I've been searching around on this forum for different threads on pike, halberd and general fighting mechanics and found this to be an oddity i havent seen discussed.

So, what gives? Could it be that different units have some hidden stat that determines their collision? Could that also be the reason why generals bodyguards are so ridiculously good charging even with relatively "bad" charge bonuses(some of the factions have iirc 5 charge bonus on their generals). If so, what stat would that be?

[FactionHeir]

As you realized yourself, thread necromancy is bad and in violation of forum rules for the Citadel.

And I think that your observation has little to do with attack value personally and more with unit/mount mass.