One quick AI fix which would make things MUCH more interesting.. (Full Version)

All Forums >> [Current Games From Matrix.] >> [World War II] >> Steel Panthers World At War & Mega Campaigns



Message


Khan7 -> One quick AI fix which would make things MUCH more interesting.. (1/6/2002 3:24:00 AM)

..or at least quite a bit harder. Don't get me wrong, it is very *INTERESTING* to have the AI send it's unsupported tanks hurtling onto my well-prepared defensive line, and have an infantry horde lumber on up behind once it's almost annihilated, but it's kind of ridiculous. And this doesn't just go for assaults, but all situations. I've had two battles in a row now where the AI barely used its infantry and threw it's tanks into the shredder. It would seem that a simple fix to have the AI adopt a bit more of a feasible combined-arms approach, or at least a "infantry = grunt therefore send them in first" approach, would make the AI an enormously more challenging and realistic opponent. The masses of German infantry about to march up through the massed ranks of their own burning tanks will most certainly put the final clinch on my poor limeys' demise; but the thing is that they could've done the same thing, with a small fraction of the losses, using the same AI engine-- if they had just sent the grunts in FIRST (is this not what grunts are for??). Matt P.S.: Ak, almost forgot: I figure it's good form to mention a positive aspect with every criticism one makes of the game. So here goes: in the battle I'm currently playing, I was quite pleased with the depth and extent of the SPWAW damage system for tanks. Perhaps a bit annoyed by it, but pleased nonetheless. Working with early '40 tanks and guns, I'm finding that about 1 in every 10 penetrating hits actually does jack squat other than make an ugly hole. And then there are so many different possibilities for that 1 in 10.. impressive. Annoying for sure, often blasting 12 holes in a tank before KO'ing it, don't know WHAT we'd do without premature bailouts.. but it seems quite realistic. I like it. [ January 05, 2002: Message edited by: Khan7 ]





Warrior -> (1/6/2002 3:49:00 AM)

In most scenarios I design, I cut the AI armor speed to that of the infantry walking speed. Doesn't make the AI any smarter, though.
You can probably edit tank speed in your Oob's. [ January 05, 2002: Message edited by: Warrior ]





Charles2222 -> (1/6/2002 4:01:00 AM)

Khan7: Hey, get 70mm guns and fire them on size three or less targets and you'll see destruction takes 2, maybe 3 hits. Peashooters don't cut it. Funny though, it seems as though the smallest gun used for AP purposes, that being the ATR, has a FAR deadly kill rate then the 20mm and 37mm guns early on.




Paul Vebber -> (1/6/2002 9:11:00 AM)

A friend of mine who is a USMC LTCOL told me when I figure out the AI that does basic combined arms tatics well to let him know so he can trade his CAPTs in for it Then there is the problem that most combatants did not use good combined arms tactics consistantly throughout the war (even the Germans, but they teneded to be better than most. THe British use Tank Squadroins en masse late into 44 nad the Russians treated tanks like doritos. "Crunch all you like, we'll make more" which was why german units outnumbers 5 or even 7 to one could retire in good order (or take 6 of the 7 to one with them going down...)




Alexandra -> (1/6/2002 9:23:00 AM)

Paul's right. For the most part, combined arms is a *post* WW2 doctrine. Very few nations actually used it in the war. Gamers use it more than thier real world counterparts do - in fact, most WW2 head to head games - SP included - tend to be '80s tactics with '40s weapons. It would be real cool to find a way to design an engine that would make players play WW2 the way it was really fought Alex




Dogfish -> (1/6/2002 9:30:00 AM)

Aha! Quick Fix... Another fine oxymoron for my collection.




Khan7 -> (1/6/2002 10:13:00 AM)

How very true indeed, Paul and Alex, but I would really have to insist that in this case the AI's antics are a completely different order of magnitude. Also it would not seem that actual commanders could have been THAT wrong THAT much of the time, as extreme errors have a way of being corrected naturally. To be thorough, I will cite some actual examples. Off the top of my head, I can cite two examples. In the British breakout at El Alamein, infantry led the charge, and tanks were not involved at all until late in the battle. In Operation Cobra, the breakout from Normandy, infantry was used in the same way. Plus, it would just seem that using infantry for the more dangerous tasks would be the logical, easier thing to do, as opposed to the other way around. Everyone had enormous masses of infantry, and realized their importance, so why would they not use them? Now, what we're talking about here is not bad coordination, bad communication, bad timing, confusion.. we're talking about the tanks go in ALONE, and the infantry shows up lumbering along WAYY too late. I cannot recall a single instance that I have heard of in which an offensive on a heavily fortified, static line, was made with totally unsupported tanks. I'm sure it happened, but it would not seem as if it could have been the norm. In the game as it is, it is invariable, from what I've seen, and according to other players' testimonies. However ill-advised much of WW2 tactical doctrine/training was throughout much of the war, I think a comparison of AI antics to actuality is truly a gross comparison, and kind of unnecessary given the mountain of justifications you guys have for not moving on this little issue. However, it would also seem to me actually that this is a bit of an "easy" fix, relatively speaking, as it would just involve getting the tanks to slow a bit, or the infantry to speed a bit, and even a very messy job of it would greatly improve AI effectiveness. I suppose if you wanted to try and balance it well and implement adjustments for tactical doctrine etc., it would be quite a task, but I would point out that currently it is not balanced, and currently there is no simulation of tactical doctrine, so it's really safe and clean just to give a great boost to the AI's effectiveness, narrow-sighted job, see? I mean, the AI would still be playing a few cards short of a deck, which would fit your bad coordination/tactical doctrine thing, but at least it would have one more thing going for it. Seems to be a more healthy way of handling it than simply upping the AI buy points or stats or whatever. Anyway, in case I didn't mention before, it's just feedback, pardon me if it appeared to be anything more. Matt [ January 05, 2002: Message edited by: Khan7 ]





tracer -> (1/6/2002 12:08:00 PM)

I noticed that the AI positions its tanks well behind the infantry; kind of giving the grunts a 'head start'. But the tanks always overtake the infantry before they reach the enemies lines. Maybe coupling these more distant start points with Warrior's idea of cranking down their speed a notch might work.




Dogfish -> (1/6/2002 2:59:00 PM)

Just a thought (untested), maybe even a quick fix. It would seem that if you want the AI to use a combined arms force when attacking then one should create combined arms formations. 5 tanks and 5 infantry per formation. Wouldn't they stay grouped and arrive at same time? This formation can then be nested into another formation, an attack force, of 10 of the smaller formations.




skukko -> (1/6/2002 5:24:00 PM)

So, as a sort of scenario -designer I would like to have 'glide-button' into editor where I can adjust easily and softly overall speed of AIs vehicle movement. Takers? maybe into CL ? Finns did attack alot without armor support... Do yourself a favour and order Lost Victories. Then you know what can be done with spwaw-engine. mosh




RockinHarry -> (1/7/2002 12:19:00 AM)

quote:

Originally posted by skukko:
So, as a sort of scenario -designer I would like to have 'glide-button' into editor where I can adjust easily and softly overall speed of AIs vehicle movement. Takers? maybe into CL ? Finns did attack alot without armor support... Do yourself a favour and order Lost Victories. Then you know what can be done with spwaw-engine. mosh
Yeah...would be nice feature!...or
if someone let us know, where the info on vehicle breakdown and other damage is stored in the DAT file!?? I think editing vehicles to "suspension damged" is more reliable, than simply cutting down the speed in the D-key edit screen!? Most of this "slowdown" edits will usually be repaired by the AI within the first 3-4 turns from my experience. Darn...the data must be somewhere..
Now checking with Hex-Edit and file compare function... ________
Harry




RockinHarry -> (1/9/2002 1:35:00 AM)

quote:

Originally posted by RockinHarry:
Yeah...would be nice feature!...or
if someone let us know, where the info on vehicle breakdown and other damage is stored in the DAT file!?? I think editing vehicles to "suspension damged" is more reliable, than simply cutting down the speed in the D-key edit screen!? Most of this "slowdown" edits will usually be repaired by the AI within the first 3-4 turns from my experience. Darn...the data must be somewhere..
Now checking with Hex-Edit and file compare function... ________
Harry

quoting myself a little bit Well..seems I erred with the new speed settings repaired during the course of the game. Testing these with official V7.0 I noticed that does not happen any more?!
Does that mean that is a "hidden" fix in V7.0?
Would be really great, if so! Opens lots of opportunities for scenario designers to get the AI a much better timing in various things. Somedody else experienced that edited speed settings now stick in official V7.0??
(i tried speed settings with 9 and 18, which seem to be very frequent for vehicles with "damaged suspension" f.e.) _______
Harry




Nikademus -> (1/9/2002 2:56:00 AM)

we should be thankful too, that the AI no longer charges in blindly, leading with it's halftracks anymore. That was by far the worst AI fobile, especially for SP-2 I observe that the AI tends to cautiously dismount it's halftracks (escorted by tanks) just before my start lines in an almost careful looking way. almost fools me into thinking i'm playing a sentient player Course it all goes to pot once the infantry start legging it, the Tanks overtake them at this point. If you have mines.....ug, it's almost unfair. AI has no clue about mines. Maybe it figures that it can blast a corridor through if only it sends in enough tank hulls




Paul Vebber -> (1/9/2002 3:36:00 AM)

THere are many "oh by the way" one line fixes that get in but I fail to document I'm pretty sure MPs and victory point changes "sticking" is one of them, but I don't we did anything that was not already there to deter friendlies from dropping bombs on friendlies, I'm certain I've seen that happen occasionally for a long time... [ January 08, 2002: Message edited by: Paul Vebber ]





Grumble -> (1/9/2002 5:43:00 AM)

Khan,
In Operations Goodwood and Totalize, after action reports on the Commonwealth side ROUTINELY decried the lack of coordination between infantry and armor. Armor generally refused to go in unless the infantry went first and vice versa. As Alexandra noted, combined arms is really a post-WW2 concept. American cooperation was somewhat better, but really was unit by unit
"Tank Tactics" a recently released book by a Canadian Author goes into some detail about this, as does "Steel Inferno".




Khan7 -> (1/9/2002 10:33:00 AM)

How right you are, Grumble, that combined arms tactics, like most things, were a learn-as-you-go thing in WW2. This is, however, not nearly the point. Shying away from grandiose predictions on just how much improvement the AI could use, I will simply say, in total security, that if something were done about this particular issue, and if it were done *perfectly*, the competency of the AI would remain *safely* below that of *any* WW2 commander above the rank of Corporal. Simply put, there is precious little risk of making the AI *too good*, which is basically what you are warning against. I hope that clears things up a bit, respectfully, Matt




MacCready -> (1/9/2002 10:42:00 AM)

Yes AI Needs Improvement,Far to predictable.
Never does anything unexpected. I don't however fault Matrix,but I do hope that the Combat Leader AI will be fully adjustable as far as how it plays the game. Various stances would be nice like 1. Agressive
2. Defensive
3. Tactical
Maybe various degrees of each could be added to create correct AI mix. And the real kicker would be simulate last human opponent posibility?




Page: [1]

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
1.453125