Archery range bug

[TosaInu]

What is a factor in all TW games? Randomness. Tactics play a role of course, but it's also a matter of a dice. I know people will jump up in frustration now, 'I'm l33t! It's my skillz(TM)': relax and read.

We've most likely all run into this phenomena: you have a tough unit and it's going to whipe a lesser enemy unit. Several conditions are in your favour, but your tough unit still loses. The frequently used routine is to accuse the other of cheating then. There are exploits in TW games and perhaps also cheats: but there's luck too.

Another very evident situation was RTW 1.0. Both teams saw their army being victorious and you could only find out when chatting with each other about the battle (disable the chat and everyone would feel great ).
Replays from the totally different battle were 99.99% identical, only one part at the beginning was different: the value of the dice.

Combatcalculations are using stats, tactics (flanked, backstabbed, extra bonusses), but also a seed (dice). When two identical units A and B fight each other under equal conditions, everything can happen. Big victory for A, big victory for B or anything in between. The same is true for any combination of A and B, though it's more likely that the stronger unit will win. And when A is much stronger than B, A will almost always win. That's tested over and over.

I am not sure about completely removing the luck factor -we can see some realism in that, a soldier slipping in mud, a salty sweatdrop dripping into an eye-, but TW is not a tactical game anymore when the luck becomes too important.

When you fail to rout, it's because your tactic wasn't good enough, your unitselection has to be improved or just that the dice was not in your favour. Use CTRL A CTRL O next time.

[-Silent-Pariya]

When you fail to rout, it's because your tactic wasn't good enough, your unitselection has to be improved or just that the dice was not in your favour. Use CTRL A CTRL O next time.

Your wisdom astounds me

[Puzz3D]

This is also true in MTW. The shield provides added protection from projectiles to the front and left. We measured this in tests in MTW, and it works. I've learned the hard way that you have to test features to make sure they are working.

This isn't correct. CBR and I ran tests last night in MTW/VI v2.01 on a flat map with no rain and no fatigue, and found that shields add protection to the front. For 8 archers (60 men, 28 arrows) shooting at 8 chiv sergeants (100 men in 10x10 blocks, armor 3 + shield 2) at a distance of 80 meters, total kills out of 800 men were:

183 front
466 left side
495 right side
433 rear

Another test suggests that the shield protection covers an arc somewhat less than 180 degrees. CBR found a post by LongJohn that supports these results.

Troops shot from behind count as having lower armour, if they have shields (as their armour includes the effect of the shield ). The game doesn't look at whether a unit is being shot from the unshielded flank, because of the graphical limitations that mean that the shield can appear on either flank depending on which direction you look from.

Troops who use two handed weapons don't get the benefit of their shields while fighting.

We also did a test of 8 archers firing at a unit which doesn't have shields, 8 swiss armored pikemen (96 men, armor 4), 6 ranks deep at a distance of 80 meters which shows no bonus for rear shooting. Total kills out of 768 men were:

389 front
372 back

[Puzz3D]

What should be the factor which would decide this? Once more this is theoreticaly asked question. This is what I'm asking. For example this happens many times in chess where match can result into the draw. I've asked hypothetical question.

If everything is equal, randomness decides which unit wins. This is why you don't want too much randomness in the combat results. The winning unit shouldn't win by a lot if they are both equal and used correctly. You reduce randomness by increasing the number of combat cycles that it takes to resolve the combat. The old engine has a 1 second combat cycle and with 60 man units the randomness is suitably low which allows the tactical use of the unit to better influence the outcome. For instance in the old engine, typically a 20% advantage (1 combat point) in combat power will produce 6 wins out of 10 all other things being equal.

[Carl]

The chance of hitting should taper off at longer ranges if the archer is shooting at a specific man. This automatically happens in the old engine because accuracy is a small error added to the trajectory of the arrow so the further away the target man the larger the positional error of the arrow when it reaches the target.

This is pretty much what I just described the M2TW system as, I just used less technical language, let me answer a few more of your points and I'll come back and re-explain it to you using some of your terminology now that I know it.

Loss of kinetic energy due to drag isn't modeled, so that makes all projectiles that hit at longer ranges have a greater penetrating power than they should have. The continual reduction in chance to hit at longer ranges in the old engine tends to offset the non-declining penetrating power, but of course they don't exactly offset in such a way to give the same result as a more realistic ballistic model.

Actually this feature has been put back in for M2TW, arrows that come in flat have a higher chance of getting a kill than one that comes in from above. The effect isn't that noticeable with arrows, (although it is there), but it's much more noticeable with Crossbows and Muskets.

The old model seems better here since the arrow is fired at a 45 degree angle at the open fire range. This means they can't shoot any further than their open fire range on flat ground. That range is described as the weapon's effective range, but it's less than the weapon's true maximum range. LongJohn explained that the effective range was chosen to be less than true maximum range so that the battles would have a larger and more epic feel to them.

Red Harvest did tests on arrow velocity in RTW, and his conclusion, as I remember, was that the arrow velocity was unrealistically high. It's definitely much higher than it was in the old engine. This is why you get those extremely high arc trajectories when the low trajectory is blocked by a structure.

Well the velocity is much closer to what it that in M2TW, my tests show that arrow velocity is just high enough in M2TW for the arrows to reach 180, (the range of muskets, the best bows are 160 for reference). So they still fire at less than 180 when firing at their maximum listed range, but only just, (you can't tell visually).

Each archer aims at a man. If the man moves, he won't be on the spot at which the arrow was fired when it arrives. If the target is far away, the arrow will miss even if the accuracy error was zero. You will get some secondary hits on men that were not the primary target. You can also get secondary hits on stationary targets if the accuracy error on the arrow's trajectory is large enough. MTW introduced the lethality parameter, and it's much easier to control the effectiveness of projectiles with that parameter rather than using the accuracy parameter. Also, LongJohn did say that he made a change in VI that should improve the archers ability to hit moving targets, but he never said what the change was.

What you've just described is very similar to how things work in M2TW. I'll give you that explanation know to help explain what I mean:

The first thing I need to go into is to mention that their are 2 types of Positional Error.

Type 1: This is used by projectiles that DO NOT have an absolute fixed Accuracy value. Arrows and Crossbows use this kind of offset. What happens here is that the accuracy value for the projectile is consulted and it is then determined from this weather an arrow will hit or miss. If it is determined to be a miss then a positional offset sufficient to cause the arrow to miss both the targeted man and the unit he is part of. However the value used is the one that would be used if the weapon was firing at a 45 degree angle. At shorter ranges where the angle into the air is less, the Positional Offset may not be high enough to cause the arrow to miss the target, at which point you still get a hit and it still counts as a hit, just as it did in the older engine. (Anything without an accuracy value listed in it's Descr_Projectile entry will use this type)

Type 2 Positional Offset: This is identical to Type 1 with the exception that the offset is calculated to match the angle being fired at, so regardless of the range between the target and firer the arrow will miss is the accuracy calculation says it should. All Gunpowder and Art weapons use this. (As does anything else with an accuracy value listed in it's Projectile entry).

Fixed Accuracy: Just to explain this a bit more clearly. Absolute Accuracy is when a projectile has an accuracy value assigned in it's Projectile entry. It is as the name suggests absolute and if it says the projectile will miss then it will. It is simply impossibbile for the projectile to get secondary hits against the target unit. (as detailed under Type 2 Positional Offset).

Non-Absolute Accuracy uses a combination of the accuracy values and (at angles below 45 degree), secondary hits from shots whose positional offset was not sufficient to cause them to miss at the current firing angle.

Position: This is where the real difference between the two systems seems to be. The old system probably used a system similar to what I described for Non-Absolute accuracy and it's attendant type 1 positional offset.

However, in the old engine, (according to you, I can't confirm I'm afraid), the target position was always the point occupied by the man at the time of firing.

In the new engine however, it is the target man's current point, or the positional offset point relative to the CURRENT position of the target man. Thus no matter how the man moves, an offset of zero will always cause a hit and an offset of 5 meters to the east will always cause the arrow to land 5 meters east of the targets current postion.

The purpose of this is pretty clear to me. It ensure that no matter how much someone may move, the absolute minimum number of hits never changes. No matter what at least some will hit the target. They probably did it to stop people wasting someones arrows by running into and out of missile range all the time.

Now that i understand how the old system did it, it looks a lot like it uses the old system, but with positional tracking added on.

[Puzz3D]

Position: This is where the real difference between the two systems seems to be. The old system probably used a system similar to what I described for Non-Absolute accuracy and it's attendant type 1 positional offset.

However, in the old engine, (according to you, I can't confirm I'm afraid), the target position was always the point occupied by the man at the time of firing.

In the new engine however, it is the target man's current point, or the positional offset point relative to the CURRENT position of the target man. Thus no matter how the man moves, an offset of zero will always cause a hit and an offset of 5 meters to the east will always cause the arrow to land 5 meters east of the targets current postion.

The purpose of this is pretty clear to me. It ensure that no matter how much someone may move, the absolute minimum number of hits never changes. No matter what at least some will hit the target. They probably did it to stop people wasting someones arrows by running into and out of missile range all the time.

Now that i understand how the old system did it, it looks a lot like it uses the old system, but with positional tracking added on.

In the old system, some volleys get no hits. That's because the error due to the accuracy of the weapon is added to the trajectory of each projectile at the time the projectile is fired, but the hit or miss isn't determined until the projectile arrives at the target location. In this system, it's harder to hit a moving target, and that's realistic.

The accuracy parameter is a weighting factor applied to a random number probably linearly distributed in the range of 0 to 1 for every individual projectile. Also, the target has a size. Horses are bigger than men, and therefore the chance of hitting a horse is higher. You can clearly see this effect in tests.

If the archers in M2TW can only shoot an additional 20 meters beyond their open fire range, that isn't much of an issue. However, it could be exploited because a shooted can advance say 5 meters and force a non-shooting enemy unit to retreat 20 meters. This isn't intuitive. Intuitively, you'd expect that you could retreat the same distance that the shooter advanced in order to get back out of range.

[Fenix7]

If everything is equal, randomness decides which unit wins.

This is what I wanted to read. Atm we can't avoid this in RTS games. Thought I'm convinced that even RTS games could be programmed with AI scripting aproache like it was used in 'half-life' for example if there would be will for this. For the start I would be glad if spears and pikes would be able to stop cavalry.

[Puzz3D]

This is what I wanted to read. Atm we can't avoid this in RTS games.

Some randomness is better than completely deterministic gameplay, but it makes a difference how much randomness is present. Too much isn't good.

[Papewaio]

For instance in the old engine, typically a 20% advantage (1 combat point) in combat power will produce 6 wins out of 10 all other things being equal.

Puzz3D-sama.

IMDHO this lowly dog believes that 6:4 is a 50% advantage.

6 wins out of 11 would be a 20% advantage.

Puzz3D-sama.

[Papewaio]

Your wisdom astounds me

Dear Mr S-P,



We love you too.

And so do these guys:

[Hosakawa Tito]

Your wisdom astounds me

Disagreeing is fine, being disrespectful is not. You got mail.

[Fenix7]

Some randomness is better than completely deterministic gameplay, but it makes a difference how much randomness is present. Too much isn't good.

I agree with this. Thought I'm wondering how MTW2 would feel like when spear/pike/cav bug is removed.

This archer bug doesn't bother me at all when compared to alan cavalry repeated charge bug.


p.s.

Disagreeing is fine, being disrespectful is not.

Good work Hosakawa Tito. Keep it up.

[Puzz3D]

[B]IMDHO this lowly dog believes that 6:4 is a 50% advantage.

A 20% advantage in combat power means you will kill 1.2 men for every 1 you loose. After 5 battles the stronger unit will have killed 1 whole unit more than it lost. Statistically, that should result in 1 extra win every 5 battles; i.e. 3 out of 5 which is the same ratio as 6 out of 10. If there was no uncertainty in the fighting, the stronger unit would win every time.

A 40% advantage in combat power (2 combat points) produces 9 wins out of 10. Ganging up of multiple men on single men is more of a factor now, and it shifts the outcome more in favor of the stronger unit. The weaker unit is only able to win 1 out of 10 battles rather than the 2 out of 10 it would get if linearly related to combat power.

[TosaInu]

For the start I would be glad if spears and pikes would be able to stop cavalry.

Hello Aonar,

That could be the result of a few things.

-Either luck is too important now, resulting in more of a 50-50 situation.
-The parameters are not well balanced, i.e.: spears not having enough anti-cav bonus.
-A parameter does not work long enough/at all(i.e. the spear anti-cav bonus) or a parameter does work/too long while it shouldn't (i.e. full frontal cav charge against pikes, while the pikes are supposed to be able to cancel some charge).

[Lusted]

Or, as is looking most likely, its the result of the shield bug in the case of spearmen. Pikemen are able to beat cavalry if the pikemen are standing still and the cav charge form the front. If the sshield bug gets fixed it should balance the whole spear v cav dynamic a lot better.

[Paolai]

from the list that CA posted, the shield bug will be not fixed in the next patch.

[TosaInu]

Or, as is looking most likely, its the result of the shield bug in the case of spearmen.

-Is a parameter that does not work if that's the case.

Pikemen are able to beat cavalry if the pikemen are standing still and the cav charge form the front.

That's not what is reported. The involved parameter does only work in some cases, it can be dodged. Again a parameter that does not work.

[R'as al Ghul]

I understand you very well your previous post already. It just seems that I was not specific enough.

I'm sorry for the misunderstanding, my friend.
Sometimes I wish we could all meet face to face, communication would be so much easier. However, in the meantime Puzz and Tosa answered your question and I can only agree. There needs to be a small random element. This element must not be too strong, though. Unit matchups still need to be predictable for tactics to work.

On the matter of Spear vs. Cav:
- a major influence on this matchup has the shield bug, as pointed out. However, if you take away the shield factor and put half in defense, half in Armour (as suggested by Jerome and known as the Shield fix) the spears work considerably better, but the Cav is still able to destroy the spearmen with a clean controlled charge. Either the spearmen don't have enough anticav power (they have AC=8 and 12 is max) or the mass of the horses is too high. The masses are actually (close to) realistic and if maintained will always enable a heavy (Inf or Cav) unit to push away a light unit. If one increases the value for AC in Spearmen then I assume that it will only enable them to kill Cav faster, not withstand a charge.

- Pikes actually can withstand a charge (when charged, not when clicked behind). The only difference to spearmen statwise is the entry "long pike" and the "phalanx" formation. I'd guess that the anticav mechanic that works here is in the code for the "phalanx" formation and not connected to the "pike" entry. I'm willing to test that and post my results.

[Puzz3D]

- Pikes actually can withstand a charge (when charged, not when clicked behind).

I think the shield bug will almost certainly be fixed in the patch, but the consequence of click behind is probably much harder to address. Unfortunately, you can't stop players from using the click behind once they learn about it.

I would say that cancellation of some of the cav charge is necessary for spears to perform well. In RTW v1.5, phalanx actually reverses the cav's charge bonus and it gets used against the cav itself, but that ability isn't given to spears as far as I can tell.

[Papewaio]

Be back... just need the stats tables.