PZIVB AP Use

[Mobhack]

It will fire HE if it thinks it has no particular chance against that armoured target - which includes angle off (class 1 armoured target angled at 45 degrees to the firer, has > 1 effective armour).

Best AP at 3000 metres is 1, average is only 0. However the HE ammo average is 1 - and on the odd time it can do up to 5. Even though HE needs more of an over-penetration to reliably do damage to armoured targets (1 overpen with AP has much more chance than 1 overpen with HE, unlike in the original game - we have devalued HE pen), its a better choice than the anaemic AP round. HE has a chance to de-track the enemy, same as AP, and firing that reserves your precious AP for closer in shots that may be more effective.

The 75L24 has a range of 60 hexes (both HE (its primary purpose) and AP - which is not, with a little stubby cannon. The 75L24 is for support tanks to duff up enemy infantry with, and that is why the gun has a 60 range - for long range HE throwing at soft & squishy targets. Not for long range AT fire!. The Panzer 3s are the anti-tank animal at that time frame.

The 75L24 is only worthwhile really close in as an AP round - out to to 300 it will average a 5(7 best), to 700 4(6 best), to 750-900 3(6), beyond 900 it averages 2(5)AP, and then drops to 1(2) at 1850, 0(1) at 3250 .

The code makes the choice on the average expected AP value, not the 'best' which will only arise in well under 10% of hits - the add on comes mainly from WH size, as large shells are assumed to carry more energy further. HEAT, when it arrives is average of 5(10 best) right out to maximum range. Averages are based on a sample of 5000 hits.

The 75L43, to give an example of a proper long AT gun of the same calibre - averages 12(14 best) out to 350, and at 3500 it is still a useful device with 3(4 best), so it will usually fire at little tin cans with AP at maximum range. (HE being the same AP values as for the L24).

SP has always been this way, right back to SP1. It is not going to change. If the HE rounds AP is equivalent or better than the AP at the firing range - then the HE round will be used.

Cheers
Andy

[DRG]

This question has been brought up many times in the past. Basically it boils down to the game is smarter than you think it is. The "AI" can calculate if the HE ammo, which is more numerous, has an equal or better chance to damage a tank at a given range than it's AP ammo does and if it decides it has an equal or better chance with the HE it will save the AP and fire the HE.

Don

[Charles22]

DRG said:
This question has been brought up many times in the past. Basically it boils down to the game is smarter than you think it is. The "AI" can calculate if the HE ammo, which is more numerous, has an equal or better chance to damage a tank at a given range than it's AP ammo does and if it decides it has an equal or better chance with the HE it will save the AP and fire the HE.

Don

Thanks DRG, I thought I had read something like this before, but couldn't recall it for the world. You do, however, see how in this case it's thinking is not well founded. Perhaps it's propensity to have a HE shot which penetrates at a 2, has just been terribly unfortunate in this case. I have not seen a single two hit in any of these PIZVB HE shots (at least 80 rounds fired), but maybe a couple of 1 damaging shots, if indeed it would had hit the top they might damage.

The funny thing is, I thought I was imagining things on another front. I am in a delay against the Poles, something very unfamiliar to me. When on the attack against the Poles, as usual, the short ranged AP shot (actually the computer making adjustments as you explained)of the 75L24 is hardly noticeable, but if a defensive action it hurts very badly. It just so happens an entire sector of my front's sole AFV presence was nothing but PZIVB's which made this even more profound. But the weird thing is this. The Poles act as though they're going to run you over with all that armor, and then, mysteriously, as though they saw your OOB's, they halt, or dance around, pretty much a hex out of reach of the AP shot. They did the same thing with the 37L48's on the PZ38's I have elsewhere. They know when that AP shot is going to go just to the level where it gets that one more point of damage and will stay beyond it (though in hte PZIVB case it as a matter of avoiding the AP altogether). It's really quite hard to believe. It causes something of a change in tactics, particularly since standing still unlike as in a assault battle, gives me no defensive advantage, to then advance at least one or two of the tanks up so that some penetrating shots can hit the front units. Pretty wild.

Though I don't like them waiting just beyond the AP shot like that, I have seen some other AI behavior which is quite unbelievable, but very good nonetheless. For example, I had a platoon of flame thrower engineers guarding what has become the central part of his attack, in some woods. I had a spread where there was only one engineer that would likely face the enemy alone. He destroyed the first two units (AFV's) that tried to get in there. Not a single units has tried to go there since. Rambo and then some; he's got them scared!

One last thing. Isn't it true that the opfire flitering isn't available to the AI? That's too bad if true. I feel somewhat gamey by using it if the AI does not, but it is a great system noentheless, or at least the conceot is, as whether it's gamey or not, I am at least doing a little experimenting with it. It's so a powerful feeling to be able to tell your opfire pretty much what to do in all circumstances, which if it worked for the AI too, makes for quite a decisive edge in gameplay between this product and SPWAW. I guess at least PBEM guys will have a field day wit it.

Have you had much feedback on the filtering, does it seem to work for the human players very well?

Thanks again.

[DRG]

Charles22 said:
Thanks DRG, I thought I had read something like this before, but couldn't recall it for the world. You do, however, see how in this case it's thinking is not well founded. Perhaps it's propensity to have a HE shot which penetrates at a 2, has just been terribly unfortunate in this case. I have not seen a single two hit in any of these PIZVB HE shots (at least 80 rounds fired), but maybe a couple of 1 damaging shots, if indeed it would had hit the top they might damage.

The AI knows in less time it takes to read the first letter of the first word in this sentence that it has a better chance with that gun with HE at longer ranges against armour than AP does and it knows what the odds are on a 1000 shot cycle for every range from zero to the maximum so just because you haven't seen a result yet it maybe because you haven't fired 1000 shots.

Opfire filtering was put in for human vs Human games but left in for play against the AI because it was assumed if we didn't allow it people would complain. NO the AI doesn't use it. It doesn't use arty gold spots either so if you want to play the way the AI does don't use either.

Don

[Charles22]

DRG said:

The AI knows in less time it takes to read the first letter of the first word in this sentence that it has a better chance with that gun with HE at longer ranges against armour than AP does and it knows what the odds are on a 1000 shot cycle for every range from zero to the maximum so just because you haven't seen a result yet it maybe because you haven't fired 1000 shots.

Opfire filtering was put in for human vs Human games but left in for play against the AI because it was assumed if we didn't allow it people would complain. NO the AI doesn't use it. It doesn't use arty gold spots either so if you want to play the way the AI does don't use either.

Don

I'm not sure what yo mean by arty gold spots. You mean like roads and hills, or backs of hills? It does seem pretty random though. Just last game, I had a ammo truck parked in the woods, way out from any possible spotting, and there wasn't any aerial activity prior to what I'm fixing to describe, but also in the woods there were 2 75IG's and 2 150IG's. None of them had fired a single round, and I was jsut waiting when to use them. Then from out of nowhere the ammo truck was attacked by fighters. Now I can see that for artillery lucking out like that, because you know it's random largely, and since it doesn't just hit one hex it needn't had targeted that hex anyway, but let's just say I have never seen the AI use aerial forces at random anyway. My adjustment? Next game, same thing, only I had a, for that time anyway, invincible ammo depo there. The sort that has 8 armnor minimum.

[DRG]

Charles22 said:I'm not sure what yo mean by arty gold spots.

Artillery priority hexes that a human player is issued with at the start of the game ( the number depends on the battle type and the number of FO's he has ) and can place where ever he want.

And the AI doesn't have an arty "cheat" that can see your units. I think everybody who starts playing this game thinks that at least once. We had one guy complain the AI always targeted his units then found out he played large unit games on small maps and the AI couldn't miss where ever it dropped arty but the AI IS programed to pay attention to road junctions and areas behind hills out of sight just like a human player would and it will detect firing events like mortars just like a human player would.

Don

[Charles22]

DRG said:

Artillery priority hexes that a human player is issued with at the start of the game ( the number depends on the battle type and the number of FO's he has ) and can place where ever he want.

And the AI doesn't have an arty "cheat" that can see your units. I think everybody who starts playing this game thinks that at least once. We had one guy complain the AI always targeted his units then found out he played large unit games on small maps and the AI couldn't miss where ever it dropped arty but the AI IS programed to pay attention to road junctions and areas behind hills out of sight just like a human player would and it will detect firing events like mortars just like a human player would.

Don

Ohhhh, I wasn't aware you got more pre-strike spots for every spotter. Hmm, I could have swore I saw a set number per the type of battle on that, but perhaps that's on a spotter per battle basis and I didn't realize it.

Still, you haven't EVER seen the AI attack an unknown hex with air forces have you? It's obvious the artillery is random, though of course it didn't used to be in some SP versions. I sometimes marvel at the foolishness of the AI artillery as it may be wasting away half it's artillery on a totally unknown spot where I haven't something even remotely close, but at such a time it has plenty of units it has spotted and doesn't re-direct it's partial guessing game to focus on what it does see. That's not necessarily bad for the game overall, but does look to be flawed thinking. I don't ever concentrate all my artillery on one spot, but I almost always will focus on something that i can see. Somehow, I just can't get my head around the notion of firing on a position where I haven't the faintest idea that something is there or not. Strangely enough, I never use it for smoke either. And to top it all off I almost never run out of arty ammo, though my field-based ones would if I didn't re-supply them.

Yes, I remember back in the old days, especially with SPWAW, where I would put my 88's on hills, usually just 2-4 of them, and right away the enemy arty or air would go right after them, despite their silence to that time.

[Mobhack]

gold spot = priority target. The AI has no knowledge of these, and does not use them, even if playing a scenario where the designer has set them (unless it randomly chose an impact hex which correlated with a priority target, of course!).

It plots all artillery fire missions from scratch - it does not know about 'repeat' missions or shifting fire (there is code in there that should - I added it meself guvnor! - do it - but something else in the spaghetti seems to take priority over that).

Planes, if targeted on some hex will search around for a target. The search zone is quite large. The AI may have plotted that plane 3 turns back, on something else you did (tanks moving on a road nearby say that were in LOS to it), or just simply on a random hex in your deployment area, or as 'road rage' on your road network, or near an objective cluster.

Our AI code, unlike the SP original has no 'magic spotting' abilities, and it has no ability to 'switch' the impact zone right across the map onto a useful spotted target like the original code.

Assuming it has no more valid located targets to unload on:
- It will sometimes fire into a random hex in your deployment zone, since players usually leave their artillery park there
- It will fire on hexes you fired indirect from (smoke puffs), since a human player can do that too, I gave it that ability
- It will beat around enemy or neutral objectives
- It will beat up the road network - either in your rear zone, or on the approach roads on your half of the map
- it will target hexes you fired direct from, or even Z fired from (if you do the 'hill dance' of popping up to fire and retreating back behind the crest - expect some arty presents, if it has some unassigned batteries. Popping smoke grenades is also an 'AI interest' item now - since you did that it must have ben important - so I'll send presents.
- Like a human player, it now knows about dust trails raised by movement in deserts etc

So - do not park arty by roads in your rear zone, or near objective hexes, And if you fire, relocate after a move or 3 of fire as it is healthier. If you pop smoke, clear the zone.

I have plenty of times seen AI air strikes on parts of the map where there was absolutely nobody around - but I usually don't 'shotgun' formations all over the map and concentrate my troops into the deep South/North as a lump, or half a lump North and half south with a few scouts observing the middle etc. (Which can be a pain in MBT if he suddenly unleashes MLRS or several CBU air strikes, or in WW2 when the nasty Nazi unloads 18 odds Nebelwerfers and killed an entire leg company of veteran/elite leg grunts - AARGH! but life is a terminal STD )

Cheers
Andy

[Charles22]

mobhack:

I have never seen anything like that, and there is no AI code which "measures danger" like that.

Wow, that is fantastic, because it's happened on two totally different parts of the map. Both the exact situation, where I have a few platoons of the "very same" tank overlooking a valley. The tanks in both cases had absolutely nothing obstructing them, as I couldn't destroy them. It finally got to the point that I got frustrated and brought some help from other places, but one of them is still ongoing now. This is occuring over at least 3-4 (more like 6 or 7) turns, so it's not like they haven't had time to get re-directed and take another path. Naturally, I can only tell you what it is doing, not why. Now since this seesm to be unbelievable, I will be on watch for it in the future and do a save when I have clearcut evidence, but as my saves stand, only the PZ38 one has any semblance of showing this behavior, and at this point an unbeliever wouldn't be too convinced with what it looks like now, as I have made some "other" adjustments as I said.

Again- they don't do that. However if your unit is now known about then they may be going somewhere else, or an objective nearby has flipped, or they have decided the hex entry cost is too much (with wrecks in it now) and have gone around another route. Or are simply hanging back as there is now a random choice if enemy is known, to close - the original SP 'tin lemming' code simply charged the objectives with anything that could move, regardless. Our AI code has a little subtlety built in.

They may not be doing that, but it looks an awfully lot like it learned from the first tank going down to both fires from those hills: "They destroyed us at the next hex, therefore we shall not approach more closely". I'm not kidding. It's like that engineer thing I mentioned. I haven't played the latest version of winspww2 very much but I have never seen anything like this before. See also the peculiar air strike I mentioned. There was absolutely no reason that should has happened, unless you think that the air strikes are willing to go after something at random, which I have certainly not ever seen, esecially since there was absolutely no way he spotted the units. The only thing I can figure is there's some very rare random code where they learn where a unit is or more likely just some glitch that exposed them but there wasn't any gamewise explanation for.

You will notice one key ingredient to my story of the tansk not going any further though...they both are facing hills with only one kind of tank on them. One hill has "nothing" but PZIVB's, while the other nothing but PZ38t's. So what may be happening, or at least it alwasy was in my prior games, I always had something of a mix on those hills. So while the gang could rely, in this case of the return fire being almost exactly the same, though at somewhat different distances, in the other example he could not, because they can't account for both the distance and the difference in tanks being a consistent pattern. the "only" way i have managed to break these things up at all, apart than some of the help I brought from other places in various forms, was to actually move some of the tanks closer. It's like they accounted for the original fire and it's distance, but when I uncharacteristically moved forward against heavy odds they didn't account for it, though they might fruitlessly return fire. Part of the key also, might be, that I only sent at most 2-3 of the total 10 or so tanks forward and kept all tanks firing. To leave some of them further back firing as well, though fruitless pretty much, might had kept him from accounting for the forward moving fire. Yes, I know, you say that doesn't happen, they don't re-calculate and "know" when that AP will start firing, or that it's effect, in the case of the PZ38's, will climb one more point in damage, therefore endangering them, but it's screwy nonetheless. Certainly nothing I have ever seen before.

Good to see the stories about the filtering as well.

[Mobhack]

Charles22 said:

Thanks DRG, I thought I had read something like this before, but couldn't recall it for the world. You do, however, see how in this case it's thinking is not well founded. Perhaps it's propensity to have a HE shot which penetrates at a 2, has just been terribly unfortunate in this case. I have not seen a single two hit in any of these PIZVB HE shots (at least 80 rounds fired), but maybe a couple of 1 damaging shots, if indeed it would had hit the top they might damage.

the sample in the hack bit of code I have, measures five (5) thousand actual hits - and generates an average result, rounded to integer. Average for that gun HE, is 1, best is measured in passing - no count of whether there were 1 at 5 best, or 15. (more like 5 in 5K shots I believe though!). But at that range, the L24 average HE round gets a 1, and the AP gets 0. So a HE round goes down range.

And I have done some * damage to a few Nazi panzers of the little types (Mk1 and 2s) in my ongoing WW2 LC, a section of A9 CS tanks have happily blown more than just paint off such, and also killed little Italian tankette thingys not worth a 2 pounder round. Not very reliable though, better way to kill an italian tankette is the Ethiopian way - charge with spearmen and manually flip the thingy on its back like a tin turtle..

The funny thing is, I thought I was imagining things on another front. I am in a delay against the Poles, something very unfamiliar to me. When on the attack against the Poles, as usual, the short ranged AP shot (actually the computer making adjustments as you explained)of the 75L24 is hardly noticeable, but if a defensive action it hurts very badly. It just so happens an entire sector of my front's sole AFV presence was nothing but PZIVB's which made this even more profound. But the weird thing is this. The Poles act as though they're going to run you over with all that armor, and then, mysteriously, as though they saw your OOB's, they halt, or dance around, pretty much a hex out of reach of the AP shot. They did the same thing with the 37L48's on the PZ38's I have elsewhere. They know when that AP shot is going to go just to the level where it gets that one more point of damage and will stay beyond it (though in hte PZIVB case it as a matter of avoiding the AP altogether). It's really quite hard to believe. It causes something of a change in tactics, particularly since standing still unlike as in a assault battle, gives me no defensive advantage, to then advance at least one or two of the tanks up so that some penetrating shots can hit the front units. Pretty wild.

I have never seen anything like that, and there is no AI code which "measures danger" like that.

Likely, it was in a tizzy trying to meet some other criteria (objectives or whatever) or had passed beyond the command radius of the platoon, and was trying to get back there, if the commander was still mobile. (There is now a small attempt to close on the command unit in that case, unlike the original SP code which did not care if a platoon was splattered all over the map and so in danger of being out of CC).

Though I don't like them waiting just beyond the AP shot like that, I have seen some other AI behavior which is quite unbelievable, but very good nonetheless. For example, I had a platoon of flame thrower engineers guarding what has become the central part of his attack, in some woods. I had a spread where there was only one engineer that would likely face the enemy alone. He destroyed the first two units (AFV's) that tried to get in there. Not a single units has tried to go there since. Rambo and then some; he's got them scared!

Again- they don't do that. However if your unit is now known about then they may be going somewhere else, or an objective nearby has flipped, or they have decided the hex entry cost is too much (with wrecks in it now) and have gone around another route. Or are simply hanging back as there is now a random choice if enemy is known, to close - the original SP 'tin lemming' code simply charged the objectives with anything that could move, regardless. Our AI code has a little subtlety built in.

One last thing. Isn't it true that the opfire flitering isn't available to the AI? That's too bad if true. I feel somewhat gamey by using it if the AI does not, but it is a great system noentheless, or at least the conceot is, as whether it's gamey or not, I am at least doing a little experimenting with it. It's so a powerful feeling to be able to tell your opfire pretty much what to do in all circumstances, which if it worked for the AI too, makes for quite a decisive edge in gameplay between this product and SPWAW. I guess at least PBEM guys will have a field day wit it.

Have you had much feedback on the filtering, does it seem to work for the human players very well?

Thanks again.

The AI has its own criteria for firing opfire - which is built into the same code you have, if you let your troops fire opfire unfiltered.

The AI will use the opfire filtering - if a scenario designer has set it for the AI. Scenario designers can now set 'fire sacks' etc. This does not matter if you have the CD or not - the scenario is still playable if designed with opfire filtering, if the player is a free game user. Ditto - your units may have opfire filtering set by the scenario designer - that is why the free user has the ability to enter the opfire filtering screen, but no ability to do any setting of data apart from cancelling it, if you want to go to regular fire algorithms.

Some scenario designers have expressed a desire to make some AI defensive scenarios that utilise the feature.

In MBT - I use it often, to e.g. filter my ATGM to only go for the targets of interest (slot the T-72, but ignore the T5X and APCS and little scout cars). In both games, I find it useful to set up my tanks and ATG to ignore grunts beyond a few hexes, and shoot armoured stuff only (the accompanying infantry can opfire the soft and squishy stuff - I want my Valentines to put 6 pounder AP into the P3 and P2s only thank you very much !). Actually - if I have no scout cars and halftracks about, I usually filter out the P2s as well, they can run around and be annoying if I only have valentines, and I can use them for target practice one the Mk3s are toast.

I rarely use the circle of interest - that might suit a scenario designer making a specific kill sack - but have used it in MBT as a 'kill any armour coming out of that wood where the road exits it' tactic, when I knew the wood was full of loaded APC and a few tanks after an air strike had spotted them, and I was prepared to ignore the few tanks I knew were off to a flank as irrelevant for now.

Cheers
Andy