• Please make sure you are familiar with the forum rules. You can find them here: https://forums.tripwireinteractive.com/index.php?threads/forum-rules.2334636/
  • Weve updated the Tripwire Privacy Notice under our Policies to be clearer about our use of customer information to come in line with the EU General Data Protection Regulation (GDPR) rules that come into force today (25th May 2018). The following are highlights of our changes:


    We've incorporated the relevant concepts from the GDPR including joining the EU and Swiss Privacy Shield framework. We've added explanations for why and how Tripwire processes customer data and the types of data that we process, as well as information about your data protection rights.



    For more information about our privacy practices, please review the new Privacy Policy found here: https://tripwireinteractive.com/#/privacy-notice

Release: ArmoredBeasts 2.08

Amizaur

FNG / Fresh Meat
Aug 18, 2006
275
3
0
43
Gdansk, Poland
Moved here from "Release ArmoredBeasts 2.06 beta (update)" thread:

What do you mean by "gun hitbox" ? There is no such thing in AB (yet).

Yes the damage has strong random factor. Damage values are fixed, but actual damage done to target is anything from 10% to 150% of shell damage value (for AP, HE, PF, even Satchels).

Do you believe that in real life, you always had to hit enemy tank twice, before it exploded/started to burn ? Sometimes one penetration was enough (if you hit ammo or fuel, or shrapnels from your shell hit ammo/fuel), sometimes a shell could go trough the tank and even not hurt the crew at all (very lucky crew), usually something in between. Some shells were more powerfull and deadly than others, too.

What is not modeled (and should be) is the fact that USUALLY the first penetration, even though it didn't kill the tank (it could be perfectly operational), was enough for the crew to start evacuating. Because some or most of them were wounded, often someone was killed, everyone was shocked. And next penetrating round was very likely to arrive in seconds... so usually everyone wanted to get out of this steel trap as fast as possible after the first penetration, they didn't waited to check if the tank is ok or not (it is not going to be anyway, as enemy loader just put next round in the chamber...).

So in most cases first penetration should AT LEAST in 90% of cases disable the tank from combat for some time (shock), almost always some of crewmembers should be wounded and often someone killed. even in case of such non-fatal penetration.

As there is no crewmember death possibility currently, or any way to paralyse the tank for a while after it was hit, so the players of penetrated tank can return the fire at once and this is not realistic (be it stock RO or AB).

Curenlty in AB so far, only the chance of (fatal tank explosion/fire) OR (all crew being killed/seriously wounded) is considered at the moment (2.08) - and this is visualised as "tank exlosion". The "tank damage" code in AB is very old (from very first versions) and ment to be completly rewritten.

Tank health in AB is going to be based mainly on size of the internal space, how much fuel/ammo is inside, how well protected ect.
For example T-34 tanks had very cramped internal space (all crewmembers close to each other) and unprotected fuel tanks inside the crew compartment - any shell that passed trough them or went off close to them would turn the tank into a torch, with crew burning alive...



As for PzIV vs T-34 duel being a lottery - what would you expect ? Both tanks have almost equal chances, and such frontal duel (very stupid thing) is a lottery.

Maybe you are talking about fact, that even if you fire first and penetrate the enemy, he can be unharm fire after you and kill you ? I second you in that, it's not ok. I'll fix it when I have time, adding crewmember death possibility (hitbox and random) and I would like crew/tank paralysation too.

But in stock game it's almost same. You have to hit enemy twice to kill him (if not hit ammo box). If he reply the fire and he does hit you in ammo box - he can kill you in one shot. Not much difference.

And how it works now (2.08 with 10%....150% of base damage).

shell damage

75L48 435
75L70 500
88L56 650

T34/76 gun 630
T34/85 gun 540

tank health values as in 2.08:

PIVH 600
Panther 650
Tiger 700

T-34/76 470
T-34/85 500

So firing 75L48 against a T-34/85

shell damage 435, target health 540

min damage 43.5, max damage 652.5 , average 348

chance for 1st hit kill: 18.5%

average hits to kill: around 2 penetrations

max possible number of needed hits (target health/min damage): 13
chances that 13 shots are needed: zero
chances that enemy dies only after 5th shot: 1%

it's not statistical analysis, just basic math

statistics would be that in most of cases you need 2 hits, sometimes 1 or 3, rarely 4, very rarely (1%) 5 hits. I think it's perfectly realistic, assuming that IRL you fire at the enemy untill he either explode, starts burning or you see the crew escaping (and then you still want to kill the tank).

Results for 88L56 shell against T-34 would be different of course (most of kills is 1th shot).

The current tank healt values are very old, not reviewed from long time, and will be adjusted in AB 2.081.
I was working on tuning the damage but it was not finished when 2.08 was released, so changes were not included.
The random minimum damage is to be changed from 10% to at least 20% (maybe more, has to be tested), some shell/tank values adjusted.

it would look something like that:

tank health values AB 2.081:

PIVH 540
Panther 560
Tiger 600

T-34/76 500
T-34/85 520

but I can change something.

In 2.09 (when I have the damn time to make and test, the latter taking most of the time) there should be random/hitbox chancs to kill crewmans, and maybe (if I find out/decide how to realise this) a few seconds of "shocked" state after penetration, in which crew can not operate the tank. I would prefer to just back-out or red-out the screen for the "shocked" crewmembers, but I know little about effects so maybe I will just block the tank controls for a while... don't know.

P.S. Someone mentioned "hit detection bug" that is "noticeable" by the fact, that IS-2 hit from the side sometimes is hardly damaged by the shot.

Do I have to set a" AB Wiki" or something, to not explain same things for n-th time ?

First, the "hit detection bug" is something that happens both in stock RO and in AB mod - but in fact in AB it happens muvh more often (and I don't know why... the shell code, part that is called every frame, is unmodified, shell velocity and fundge factor is the same...).
When it happens - so when the shell hit against tank target is not detected properly (or sometimes - maybe not at all) - then there is no penetration at all (in stock game or AB<2.08). There are even no penetration anims/sounds - it looks like if the shell hit dirt under the target of flew right trough it.
And if there is no penetration, there is no damage at all !!

So when you hit an IS-2 and see it "barely smoking" after that, then don't blame the "hit detection bug" because you DID in fact damaged the target (it smokes now) so you did hit and penetrate. Everything is alright. Just damage happened to be low.

And in the end - in 2.08 the "hit detection bug" is partially handled by the code. Most of the time when it happens it should be fixed (for most of front and side hits) . And penetration/damage applied correctly then. You shouldn't even notice.
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Thanks for the in depth explanation Amizaur. My original question was in regards to close range shell effects, and if there was any logical reason to think there is a system quirk or oddity with the IS2? Or have I just had a bout of bad luck?

Also -- is 85mm AP supposed be be a shell damage value of 630, and 76mm AP 540?

shell damage

75L48 435
75L70 500
88L56 650

T34/76 gun 630
T34/85 gun 540
Jeff
 
Last edited:

Amizaur

FNG / Fresh Meat
Aug 18, 2006
275
3
0
43
Gdansk, Poland
Well, the 76mm damage is larger because AFAIK it had awfully large HE charge :) even in comparison with 85mm one. And the "shell damage" value was initially meant to reflect only "explosive" damage initially. A kinetic energy component was to be calculated "on the fly"... but it's still not done yet :p Instead I only increased some values for higher velocity guns (so 75L70 have better average damage than 75L48).



First I thought your "IS-2 problem" was about firing at IS-2 with 75L48. It could happen that 2-3 shots were needed. But now I found original post and see it was about firing at PzIV with 122L43 gun. Well... it definitely should be 1 shot kill in AB. (edit: 80% chance for 1th shot kill, but at least should be severly damaged)

So definitely hit detection bug. At short range number of "hit detection bugs" can in fact be much higher (shell velocity is higher). And, unfortunately, PzIV is a "special" tank, as (for some very strange reason) there is much, much more hit detection bugs happening against PzIV than against other tanks. Have no idea what is so different in PzIV model... :/

But it should not happen anymore in 2.08 as most of the bugs should be corrected "on the fly" by code. A clean side shot too. When it is detected that shell hit - for example - rear armor FROM INSIDE OUT (because the angle is "little strange") - then it is assumed that it was front hit in fact and the front armor rating is used instead and the angle is corrected. If same happens for the side - just angle is corrected (armor rating for both sides of tank is same anyway). By this simple method, probably about 80-90% of "hit detection bugs" can be still handled correctly.

I wish I knew why in hell the PZIV (one of PzIV models, can't remember if it's H or F2) is several times more prone to this bug than rest of tanks... :( What's so "special" in this model... knowing this could be the key to eliminating, or greatly reducing the bug...
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Sorry...was re-reading your post. I know you are working within the confines of the original game engine. My questions are not intended to be critical in any way. I enjoy the AB-Mod and appreciate your efforts in trying to inject additional "realism" into the system. I am very interested in how the tank model damage system works.

You indicated the following:

tank health values as in 2.08:

PIVH 600
Panther 650
Tiger 700

T-34/76 470
T-34/85 500
Are these frontal armor "health" values? Moreover, does each vehicle have different health values for side and rear armor?

Also, how does shot line angle affect the damage matrix? I mean, if a tank facing is angled relative to the shot line of the projectile does this influence the health value or shot damage potential...?

Thanks
Jeff
 

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Well, the 76mm damage is larger because AFAIK it had awfully large HE charge :) even in comparison with 85mm one. And the "shell damage" value was initially meant to reflect only "explosive" damage initially. A kinetic energy component was to be calculated "on the fly"... but it's still not done yet :p Instead I only increased some values for higher velocity guns (so 75L70 have better average damage than 75L48).

At short range number of "hit detection bugs" can in fact be much higher (shell velocity is higher). But it should not matter in 2.08 as most of the bugs should be corrected "on the fly" by code.

But I would rather think that (if this is what you talked about
I see. Very interesting.

Yes I agree a large bursting charge -- assuming the base fuze functions properly after plate perforation -- should in theory have greater behind armor effects. Although various folks would argue this point, citing the British design philosophy of abandoning use of bursting charges in their tank and antitank gun fired armor piercing projectiles. Even the Germans had numerous issues with base detonators functioning high order following plate perforation. But given high order detonation following perforation, it seems logical that incendiary and fragmentation effects would be enhanced over that of simple armor piercing shot.

The problem is that larger bursting charges are typically associated with lower perforation potential. This can be partly attributed to lower mass of the penetrator, but is more a function of lower strength of the penetrator due to the large void at the base of the projectile.

But, be that as it may...it is purely an aside and me meandering off on something that is not relevant to me understanding the game damage system.

Thanks for your reply.
Jeff
 
Last edited:

Amizaur

FNG / Fresh Meat
Aug 18, 2006
275
3
0
43
Gdansk, Poland
The fuse reliability was to be part of the model, and we could then discuss what value should be set for different shells :). In case fuse failed, only kinetic damage was to be applied. In "full" version of the model (rather not intended for RO) even a "fuse critical angle" is planned, for example we know that German PzGr39 rarely (or never) functioned after penetrating heavily sloped armor (like T-34 front glacis). They were effectively a solid shots then. Interestingly, it's mentioned in some raport that T-34 was very hard to set on fire frontally, but very easy from the side. Well, Pzgr39 going off inside crew compartment, with all this ammo on the floor and fuel tanks on sides... on the other hand if solid shot (non exploded Pzgr39) penetrates frontally, it won't hit anything explosive or flammable - only crew bodies - at least untill it reaches the engine compartment...

***

No, the values I have given before

(PIVH 600
Panther 650
Tiger 700)

are "tank health" values (amount of damage the whole tank can absorb, before it will explode), not armor health values.

The original RO damage model has something to do with RTS games and "energy bars" - a tank absorbs damage, become smoking ligthly, smoking heavily, on fire, then booom....

I didn't change the system to "live-dead" one at first, just reworked a bit the damage values, put a random factor on damage, and got 80% of what I needed. The "health" variable is in fact useful for "accumulating damage" - so second penetration has better chance than first one, and also health state is the only way now to set the "smoking, heavily smoking, on fire" states. Maybe later I will leave those values, but use them in different way.

I considered the "armor plate health" value - it seem the only way to simulate damage (cumulative) done by non-penetrating (and penetrating too) hits. So the only way to make something like KT vunerable, simetimes ;).
Then maybe value like square root from projectile momentum would work for "plate damage"... Not kinetical energy but momentum (V*m).

OK, have to go sleep, has to get up early :)

I'm still going to reply for your PM about plate penetration, I keep it and didn't forget about it :)

Regards!
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
No, the values I have given before

(PIVH 600
Panther 650
Tiger 700)

are "tank health" values (amount of damage the whole tank can absorb, before it will explode), not armor health values.

The original RO damage model has something to do with RTS games and "energy bars" - a tank absorbs damage, become smoking ligthly, smoking heavily, on fire, then booom....
Ok. I see. Thanks.


I'm still going to reply for your PM about plate penetration, I keep it and didn't forget about it :)
Sounds great. Looking forward to your post.
 

Demonizer

FNG / Fresh Meat
Apr 28, 2008
34
11
0
I think this is bug and not lag or anything. So anyway here goes

Both players me and this other chap had pings of alround 60-80 he was in a Tiger and i was in a T34-76 i flanked him and got behind him and was very close about 50m i aimed directly at his rear and fired the shell but it did not explode nor did it bounce of his armour but went stright thru the tank and hit the ground on the other side like it was invisible, i fired again but the same thing happened infact i fired about 6x aiming at differnt parts of the tank all shells went thru the tank like it wasnt even there. Even the other dude was shocked and said he could see the shells coming right thru the tank and hitting dirt infront of him.
 

Amizaur

FNG / Fresh Meat
Aug 18, 2006
275
3
0
43
Gdansk, Poland
I'm little shocked too :) it's very hard in practice mode to find such angle/distance that this bug occurs practically every time. And only with higher-velocity guns like from T-34/85.

You got six undetected collisions in a row from medium velocity gun in online game... hm, maybe the server was overloaded ? Have no better idea. The ping should not matter if I understand how the RO works, the projectile launch, its flight and collision detection, finally the penetration calcs and damage are all processed on server (shell flight and collision det. also simultanously on clients, but only for eye-candy effects). So it happened (most likely) on server (you didn't penetrate or damage) and also happened on both clients (your computers) as you reported seeing both the same. Hm... Same bug on three different computers.

The anti-bug mechanism I implemented works for front and side shots, I see now I have to add a code for rear shots too :eek: But, if the shells were REALLY flying COMPLETLY trough and hitting dirt on the other side - the anti-bug code would not work anyway, it only helps to correctly handle shells that went partially-trough (and hit the other side of the tank). I must admit I never seen in this game a shell that went completly trough a tank and flew away... it was always a shell hitting the opposite side of the tank that was aimed for.

Damn, why I never seen things like this in online game, so I could try and see it myself :(

I consider throwing all modified code out of the tank projectile and release such version of AB for public test, to see if the source of the problem is there... :(

edit: I checked the 2.08 sources and the bug-correcting code is used for all sides of the tank - rear armor too (I thought I was to add it later but it's already there). So if it happened in 2.08 then it really had to be case when projectiles went completly trough the tank without collision :O and continued to fly. Such thing can't be fixed by the code as projectile never interacts with the target tank... Very strange, never seen it in game (practice or online).
My question would be, on what server it happened and if the server was full at the moment it happened ? How many people played ? And what was the map ? Maybe I can catch the same map and similar server loading and do some testing with a friend ?
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Demonizer's observations do sound familiar to my own online observations -- i.e. the projectile seeming to explode behind the targeted vehicle.


However, having said that, I must admit that I am having a hard time duplicating the short range lack of projectile effect in the practice mode. I was wondering if perhaps player error in range estimation is playing some effect. I am at leisure to adjust more precisely to range while playing in practice mode. If during online gamining a player estimates range to be say 500-meters and sets the same on his gunsight reticle, but the target is really only 200-meters away, are we seeing the projectile sailing over the target and exploding behind the target. The appearance looks like the projectile passing through the target, but in fact it is going over the target.


Same again for underestimating range. I set my reticle to 200m range, but the target is 500meters away. Projectile strikes in-front of target.

Perhaps there is a mind set amongst gamers (reinforced through playing first person shooter's) that bullets travel on a straight line string between shooter and target.


Perhaps this is also partly a psychological thing -- the uncommon is always the more memorable event. Our tendency is to dwell on the uncommon as it tends to be much more interesting.
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Sorry – the above is just my own speculations\meanderings. I know enough about real world tank projectile ballistics and trajectory as a function of superelevation (initial launch angle) to understand how things should work. I don’t pretend to understand the actual trajectory modeling within the RO game engine.

Obviously superelevation for a 500m range set for a high velocity tank projectile is going to result in a higher max ordinate (a higher trajectory) relative to say a 200meter range set. However, in most real world instances – even for WWII era tank main guns – the difference in superelevation between a 200m range set and 500m range set for a high velocity tank projectile will in most cases still result in a trajectory path that will pass through the target profile (Assuming we are not talking about large ground elevation contrasts between target and firer, etc. – and assuming we were continuing to trace the projectile through the target profile to its imaginary termination point on the ground behind the target). The contrast between a superelevation required for a 500m range set relative to a 200m "actual" target range will move the MPI (mean point of impact) up higher on the target's vertical profile, but that’s why you aim center of visible mass to minimize range estimation errors on the resultant hit probability. It is the reason why “battlesight gunnery” and battlesight range settings are trained and effectively utilized. This is somewhat of a generalization as lower velocity tank guns such as the 75mmL24 used on early Stugs and PzkwIVs or 75mm used on a Sherman tank have associated hit probabilities that are much more sensitive to range estimation errors than say the 76mm gun used on a M18 Wolverine or 75mmL70 used in a Panther.

I can easily see how an in-game trajectory model is such that over-estimating range for close range events could result in the projectiles trajectory trace being above the vertical profile of the potential target. It just depends on the in-game ballistics model. And just to be clear, in my humble opinion, any semi-realistic ballistic trajectory model within games like RO are much better than the straight line LOS bullet trajectory one often encounters in first person shooter games.
 
Last edited:

[TW]Wilsonam

VP, Tripwire Int.
Oct 17, 2005
4,061
2,618
0
61
Roswell, GA
www.tripwireinteractive.com
Jeff - it may be one of those things that only happens online. In practise mode everything happens "simultaneously" on your own PC. Online, the server is managing everything and the client PCs are replicating all the effects for you. When bugs pop-up they are quite often in one mode only and are caused by wierdness in the replication.
 

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Jeff - it may be one of those things that only happens online. In practise mode everything happens "simultaneously" on your own PC. Online, the server is managing everything and the client PCs are replicating all the effects for you. When bugs pop-up they are quite often in one mode only and are caused by wierdness in the replication.
Thanks Alan. Any thoughts on why the weirdness only occurs for online short range engagements?
 

Amizaur

FNG / Fresh Meat
Aug 18, 2006
275
3
0
43
Gdansk, Poland
I checked the 2.08 sources and the bug-correcting code is used for all sides of the tank - rear armor too (I thought I was to add it later but it's already there).

So if it happened in 2.08 then it really had to be case when projectiles went completly trough the tank without collision :O and continued to fly. Such thing can't be fixed by the code as projectile never interacts with the target tank... Very strange, never seen it in game (practice or online).

My question would be, on what server it happened and if the server was full at the moment it happened ? How many people played ? And what was the map ? Maybe I can catch the same map and similar server loading and do some testing with a friend ?

And next question is - are you (Jeff and Demonizer) ABSOLJUTELY SURE that the projectiles went competly trough the target tank and continued to fly after that, untill they hit the ground ??



Note that in most cases they should fly some distance away and hit ground maybe 100m behind or even more. Only if firing from very close (like 30m) you can aim so low that projectile would hit the ground very close to the tank behind it (IF it went trough it).

I'm asking about it, because when projectile goes PARTIALLY trough the tank and hit is detected only when on the opposite side from inside out (normally it then fails to penetrate because of very odd angle value)




then VISUALLY for the player it looks VERY similar to projectile striking the ground below or just behind the target - no explosion flash, but some dirt can be visible flying in the air ect. You can notice the difference if you see it few times in controlled test, like me, but I guess when playing game you just don't look so closely on details and I didnt' notice the difference for very long too, untill I tested it.

So I'm asking if you are absolutely sure the projectile went in fact trough and continued to fly after that, untill it hit the ground (first picture) ? I know it's hard to tell as you don't see what's happening behind the target :) but maybe the other person (like in case of T-34 vs Tiger) have seen what really happened with the projectile ?

If that was in fact "normal" and well known bug so partially-trough (and hitting opposite armor from inside out, second picture), then it would just mean my bug-correcting code in 2,08 has a bug itself and doesn't work correctly. Of course I checked it but only briefly...

And in the end, I will check what is the chance of T-34/76 gun penetrating the Tiger I rear hull armor currently in AB. Armor is 80mm at some slight angle, it's set to 81mm. 76mm gun penetration is set to 85mm at 50m and little more if closer, so there should be penetrations and they should be quite lethal (T-34/76 shell has quite big damage value in AB). Penetrations against Tiger side should become probable from under 400m, have 50% probability from 200m and almost sure from under 100m. Assuming ideal angle. Yes, I remember I was to make tables/charts of those values ;P They are... in the process... still... ;)

***

The question why it happens more often in AB than in stock game is still to be answered. The velocity of projectiles is the same. I never added any code to the "tick" section of either projectile or tank, so it should not increase CPU loading while the projectile is flying or tank is driving around. Only AFTER collision is detected, the code modified in AB is executed and could marginally (single percents, measured) increase CPU loading for single frame. I also checked if the modified tracer effect doesn't cause this and of course it doesn't (effects are not executed on server anyway). I do not suspect also the single incarnation of mutator script can slow things down, after all other similar mutators are run without such strange effects. I will check this to be sure...

Have to think what AB code could possibly increase (server) CPU loading because of being executed every frame... but don't see anything really. Argh... My only candidate now is CannonPawn and DrawHUD function with modified optics code (slightly more calcs needed). I GUESS DrawHUD is not executed on the server at all (what for?) but if it was, then multiplying it by 40 players... But it probably doesn't...

Just thinking loudly... Maybe good idea would be to check stock vehicle firing at AB tank and vice versa, if the probability of the bug is changing. It would say something valuable, no matter what the result is (if the cause lies in the projectile, target tank or something other).

It shouldn't hard to do, as replicating the bug in practice mode (in AB) is not a problem, using T-34/85 firing at PzIV front from about 30-50m sometimes I can get it almost every time. It can be also repliated it in the same way in stock game (in practice mode), although it's much harder to find and almost impossible to get every time (and also hard to notice, if you don't know what to look for). At least on my computer (Athlon XP 1.9GHz, various GFX cards from very old GF2MX400 to GF 7300GT). So the difference (or lack of difference) should be noticeable...

***

O, I see the AHZ vehicles 2 beta 2 is released and ready to be converted. So I have no excuse anymore ;P
 
Last edited:

[TW]Wilsonam

VP, Tripwire Int.
Oct 17, 2005
4,061
2,618
0
61
Roswell, GA
www.tripwireinteractive.com
Jeff: I can't immediately think of anything, I have to admit - but I'm no coder. Odd ideas that spring to mind would be that you are so close that the round travels past the target within the first "tick". In 1/25th of a second, a round with m/v 1,000 m/s will move 40 meters. If some is wrong with the trace forward in the first CPU tick after firing, the round may effectively appear 40 meters away, rather than at the gun barrel - and if the target is within that range, the round would appear the wrong side of it. But I am just guessing wildly here :)
 

Amizaur

FNG / Fresh Meat
Aug 18, 2006
275
3
0
43
Gdansk, Poland
IIRC in RO the tank rounds don't start at full speed, but are spawned at slower speed and then accelerating to full speed it in first few frames to avoid such problems. I remember someone (Ramm?) explaining it somewhere.

edit: Tank shells in RO are not really 1000m/s, the parameter called SpeedFudgeScale of 0.5 makes them effectively 500m/s or less (at the same time not affecting the trajectory). So if they were spawned at 500m/s then in 1/25th of second they would travel 20m and there could be problems under 20m or less. But they are spawned at a speed of MinFudgeScale=0.025 of full speed (so 12.5m/s or less) and accelerate to SpeedFudgeScale=0.5 of max speed in InitialAccelerationTime=0.2 seconds (so 5 frames).

I have another candidate, for online mode bugs. The same shell is spawned on both client and server (on client for just being visible and generating hit effects) and for a moment they fly independently (server one can be lanuched some miliseconds before the other, because of lag), but then they are "synchronised" in some way by the game and client shell seem to be "teleported"
to the position of server shell - or something like that.

I observed this when - by accident - I had the same shell having different speeds on the server and on the client :D Very strange things happened then and client shell was re-synchronising and jumping forward and backward all the way :). If such synchronisation happened on first 50m and the client shell by accident jumped from to just behind the tank (the server shell is fired first so should be ahead) then there would be no collision on client maybe.

But this doesn't explain at all why there is no collision detected on the server... :/

P.S. Why it happens for online short range engagements ? Well one important factor may be that on short range the shells are fast, on longer range they bleed speed and are less likely to "skip" the tank in single frame... But that would not exactly explain why relatively slow 76mm shells failed to hit a large Tiger tank from close, so there are probably other reasons too.

I could tell you that in practice mode you can find a specific range (best is 50m or less) at which you get plenty of collision bugs. Then you can shorten (or lengthen) the distance by one meter or so and suddenly the collision bugs disappear... Didn't check this in online mode... but the the bug itself is present even in practice mode. In online mode there are probably other factors that can increase the number of bugs or even a different source of bugs, don't know...
 
Last edited:

ha$h

FNG / Fresh Meat
Sep 10, 2007
17
0
0
Good mod, thx for all, but IS2 is not realistic in AB!!! :mad: IS2 can't pierce Pz5 with 80mm@55⁰(~140mm effective) @ distance more than 1km, but it brake Pz5 at the seams. No ricochet!!! FG shell at turret hit destroy Pz5 turret as well. Pz4 all mod can't sustain 122mmFG shell even in front! But accuracy in IS2 was bad. Bad optics...

If you understand russian, I tell for You more.

Respect for mod!
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Thanks Alan and Thanks Amizaur for the in depths. Very insightful.

Regarding the variable projectile speeds, this got me to thinking about some of the things I do during game play. I have noticed that I can sometimes dodge projectiles -- If I see the firer or am watching the firer while I am in a moving tank. Moreover I can see the firer fire at me during game play – usually I see the initial part of flight via the tracer element or muzzle flash\smoke discharge -- than I do a quick jog to my vehicles controls -- left or right. At medium to long ranges this is usually enough to get myself out of the path of the incoming projectile. While I’ve never been shot at before, I’d guess the above is a bit far fetched from the perspective of our nebulous concepts regarding modeling realism within the game environment. I reckon arbitrary zigzagging is a realistic and effective means of reducing hit probability – but deliberately dodging incoming projectiles by watching for the tracer element seems an unlikely tactic in the real world (dodging incoming wire-guided missiles in MBTs aside). But given what Amizaur was just discussing above, I can see why it is I am able to implement this sort of thing during game play.

Best regards
Jeff
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
Good mod, thx for all, but IS2 is not realistic in AB!!! :mad: IS2 can't pierce Pz5 with 80mm@55⁰(~140mm effective) @ distance more than 1km, but it brake Pz5 at the seams. No ricochet!!! FG shell at turret hit destroy Pz5 turret as well. Pz4 all mod can't sustain 122mmFG shell even in front! But accuracy in IS2 was bad. Bad optics...

If you understand russian, I tell for You more.

Respect for mod!
Of course effective plate thickness is not 140mm -- 140mm is the LOS thickness of the plate (Line of Sight Thickness -- simple cosine relationship). Effective thickness is much more complex than a simple cosine relationship and is a function of t/d. It can also be a function of projectile bending strength, projectile nose shape, plate hardness, plate quality...blah blah blah.

Projectile effectiveness within the game is very often driven (as it should be) by side angle or aspect of the target relative to the firer. So unless the shot is straight on, invariably obliquity will be compounded as a function of side angle. When you are talking about a base plate slope of 55-degrees, effective thickness will increase rather dramatically as the true attack angle is compounded as a result of vehicle cant, side angle, etc.. Moreover the difference between plate effectiveness at say 60-degrees is rather dramatic relative to say 55-degrees.

Invariably people keenly interested in this sort of thing want to discuss firing trials and how these should be interpreted -- such as those carried out at Kubinka or the like. But this presumably has been kicked about on other threads on this forum.

Best Regards
Jeff
 
Last edited:

jeffduquette

FNG / Fresh Meat
Feb 19, 2008
339
1
0
I checked the 2.08 sources and the bug-correcting code is used for all sides of the tank - rear armor too (I thought I was to add it later but it's already there).

So if it happened in 2.08 then it really had to be case when projectiles went completly trough the tank without collision :O and continued to fly. Such thing can't be fixed by the code as projectile never interacts with the target tank... Very strange, never seen it in game (practice or online).

My question would be, on what server it happened and if the server was full at the moment it happened ? How many people played ? And what was the map ? Maybe I can catch the same map and similar server loading and do some testing with a friend ?

And next question is - are you (Jeff and Demonizer) ABSOLJUTELY SURE that the projectiles went competly trough the target tank and continued to fly after that, untill they hit the ground ??



Note that in most cases they should fly some distance away and hit ground maybe 100m behind or even more. Only if firing from very close (like 30m) you can aim so low that projectile would hit the ground very close to the tank behind it (IF it went trough it).

I'm asking about it, because when projectile goes PARTIALLY trough the tank and hit is detected only when on the opposite side from inside out (normally it then fails to penetrate because of very odd angle value)




then VISUALLY for the player it looks VERY similar to projectile striking the ground below or just behind the target - no explosion flash, but some dirt can be visible flying in the air ect. You can notice the difference if you see it few times in controlled test, like me, but I guess when playing game you just don't look so closely on details and I didnt' notice the difference for very long too, untill I tested it.

So I'm asking if you are absolutely sure the projectile went in fact trough and continued to fly after that, untill it hit the ground (first picture) ? I know it's hard to tell as you don't see what's happening behind the target :) but maybe the other person (like in case of T-34 vs Tiger) have seen what really happened with the projectile ?

If that was in fact "normal" and well known bug so partially-trough (and hitting opposite armor from inside out, second picture), then it would just mean my bug-correcting code in 2,08 has a bug itself and doesn't work correctly. Of course I checked it but only briefly...

And in the end, I will check what is the chance of T-34/76 gun penetrating the Tiger I rear hull armor currently in AB. Armor is 80mm at some slight angle, it's set to 81mm. 76mm gun penetration is set to 85mm at 50m and little more if closer, so there should be penetrations and they should be quite lethal (T-34/76 shell has quite big damage value in AB). Penetrations against Tiger side should become probable from under 400m, have 50% probability from 200m and almost sure from under 100m. Assuming ideal angle. Yes, I remember I was to make tables/charts of those values ;P They are... in the process... still... ;)

***

The question why it happens more often in AB than in stock game is still to be answered. The velocity of projectiles is the same. I never added any code to the "tick" section of either projectile or tank, so it should not increase CPU loading while the projectile is flying or tank is driving around. Only AFTER collision is detected, the code modified in AB is executed and could marginally (single percents, measured) increase CPU loading for single frame. I also checked if the modified tracer effect doesn't cause this and of course it doesn't (effects are not executed on server anyway). I do not suspect also the single incarnation of mutator script can slow things down, after all other similar mutators are run without such strange effects. I will check this to be sure...

Have to think what AB code could possibly increase (server) CPU loading because of being executed every frame... but don't see anything really. Argh... My only candidate now is CannonPawn and DrawHUD function with modified optics code (slightly more calcs needed). I GUESS DrawHUD is not executed on the server at all (what for?) but if it was, then multiplying it by 40 players... But it probably doesn't...

Just thinking loudly... Maybe good idea would be to check stock vehicle firing at AB tank and vice versa, if the probability of the bug is changing. It would say something valuable, no matter what the result is (if the cause lies in the projectile, target tank or something other).

It shouldn't hard to do, as replicating the bug in practice mode (in AB) is not a problem, using T-34/85 firing at PzIV front from about 30-50m sometimes I can get it almost every time. It can be also repliated it in the same way in stock game (in practice mode), although it's much harder to find and almost impossible to get every time (and also hard to notice, if you don't know what to look for). At least on my computer (Athlon XP 1.9GHz, various GFX cards from very old GF2MX400 to GF 7300GT). So the difference (or lack of difference) should be noticeable...

***

O, I see the AHZ vehicles 2 beta 2 is released and ready to be converted. So I have no excuse anymore ;P

Sorry Amizaur – I somehow missed your post with the diagrams.

As to your questions – no I am not absolutely sure the projectile was passing through the targeted vehicle. What I recall is that there isn’t the normal impact flash one usually sees when AP projectile strike a tank in the game. That’s how I typically know my range and line are correct for long range shoots – I see the projectile impact flash on the targeted tank (very realistic effect by the way having done a fair bit of shooting on old tank hulks up at Ft. Irwin).

What I am seeing in these very short range events is like a puff of smoke or some such thing that looks like it is either occurring on the target or behind or under the target – it just depends. Almost like the shell is disintegrating or shattering without the tell-tale impact flash of AP vs. Steel Plate. That’s why I was asking about projectile shatter modeling sometime back. The seeming lack of projectile effects on the target for these very close range shoots is the most pronounced aspect of this unidentified thingy – i.e. as the shooter I am not seeing thin grey smoke on the target like when the target vehicle has gone into a yellow damage state as a result of the shell impact, nor engine deck fire and black smoke if the target was put into a red state of damage as a result of the projectile impact. It doesn't happen all of the time -- as I say one or the other vehicle is invariably destroyed even in very short range events. It just sometimes happens that multiple hits are required. Like an IS2 hitting a PzKwIV three times at very short range before the MkIV explodes.

Regarding your rear plate comments – that is interesting. The game coding for the plates isn’t such that impact is only detected on one surface? What I mean to say is the object collision only occurs if the projectile hits the front surface of a plate?

Jeff
 
Last edited: