|
| Thread: Tactic exercices |  This thread is pages long: 1 2 3 4 5 6 · «PREV / NEXT» |
|
IRh
Famous Hero
Lizard
|
posted November 24, 2003 07:28 PM |
|
|
2 bjorn190:
Of course you're right. But, why can't you consider this as just Heroes-based puzzles? And, I'd say, there are exersices in using not BUGS (like double-casting or so), but RULES of the game. And, knowing rules, isn't it useful against any enemy? I'm sure you won't bless Angels or use StoneSkin against Behemoths, but you use 1-creature stacks for eating retals and so...
And, I never say these tactics is all you need.
2 Binabik:
Do you think vampires can take a lot of damage without hit-n-run? One Naga kick - and no vampire... OK, I'd say, just to determine: vampires AREN'T ALLOWED to attack.
With hit-n-run, you don't need your dragons at all! (but need MUCH time ).
Your split makes 8 frenzied attacks and 3 ordinary, say 20 attacks (really, less) - for about 20 Nagas, I suppose. And still 5 (no, 8!) Nagas survive.
Hint: you still need these vamps...
____________
|
|
binabik
Responsible
Legendary Hero
|
|
posted November 25, 2003 07:47 AM |
|
|
Quote:
Of course you're right. But, why can't you consider this as just Heroes-based puzzles?
Yea, they're fun.
In a real game there is always a lot going on and a lot of choices to make. With these, you focus in more on certain specific tactics and spells. I can't say I've actually learned anything new, but it's prompted me to investigate a few things more closely.
In the titan vs black dragons, near the end the draco attacked my main stack who got a retal. I had a single titan which I moved away from the main stack knowing the dracos would probably hit it next, which they did. At some higher number, the draco's would have attacked the main stack instead, but how much higher? Can that approximate number be determined without guessing?
A few maps also made me think about the hated enemy thing. Even considering the extra damage, will the hatered cause a unit to attack an enemy even if other attacks may have been better choices?
Sure, these things happen in "real" fights too, but these kind of maps bring them to the top of the mind.
Quote:
Do you think vampires can take a lot of damage without hit-n-run? One Naga kick - and no vampire...
No. I used the calculator for that at the beginning. I thought they might be able to finish 1-3 nagas at the end depending on spells points. It's just a matter of finding the right split and casting spells in the right order and *at the right time* considering how long the spells last. The short spell duration is where the vamps come in handy. I don't give up easily, and I WILL kill those nagas
I've finished all the others except Mr Ike's maps. I also went back and did ex2 in the corner instead of using terrain (19 survivors I think).
Nebuka: On ex6, are you having trouble retreating? If so, try playing around with the order of the slots you put stacks in.
____________
|
|
binabik
Responsible
Legendary Hero
|
|
posted November 25, 2003 08:05 AM |
|
|
IRh:
Is Ex9 new? The link takes me to stoker
Stoker, now THAT'S a strategy game....with high stakes and players who cheat.
____________
|
|
IRh
Famous Hero
Lizard
|
posted November 25, 2003 04:26 PM |
|
|
|
About #9: Strange... Try now please, I updated the map file...
|
|
binabik
Responsible
Legendary Hero
|
|
posted November 25, 2003 07:05 PM |
|
|
Ex9 works now.
____________
|
|
IRh
Famous Hero
Lizard
|
posted November 26, 2003 06:33 PM |
|
|
About hatred...
I't quite sure the only effect of it is an increased attack. It's not monster who decides what to strike, but a *side*. And AI is a pragmatic thing, it can only use this effect when thinking.
More, mixing , 4X, Angels and Devils won't lead to any additional consequences. Devils even will get +1 morale from Angels as well as other units.
Ans when the AI attacks larger or smaller stack - it predicts a results of each action and chooses that one taht leads to better forces *proportion*. If the retailation would lead to a severe loss, AI attacks smaller stacks, war machines etc - cowardly tactic! - but when there isn't any, it is forced to strike the main group.
But! The last isn't true for auto-combat ai - it sometimes defends in such situation to save your forces.
I remember I had 1+1 crystal Dragons with strong Mutare ang
almost lost a battle with several Hydras - my Dragons waited then defended, standing around hydras, while being attacked! Still the 1 CD damage was less than double Hydras retal damage (to two stacks) - so they decided to pass!
____________
|
|
binabik
Responsible
Legendary Hero
|
|
posted November 26, 2003 08:53 PM |
|
|
Quote:
I't quite sure the only effect of it is an increased attack
I'm pretty sure this is the case also. But I'm not 100% sure and it would be interesting to find out absolutely. Sure the AI is pragmatic, but it could have a built in hate factor. For example it goes through all the damage/threat calculations for every attack posibility and determines the AA's will do 1000 points damage to the devils. What if, for example, it adds 10% to the hated enemy number and uses 1100 points damage instead of 1000 for the final decision. I doubt this is the case, but I think it would actually be kind of cool if it were. It would be almost like adding a "personality" to the AA making them not think clearly due to their hatred.
As for the AI being cowardly, sure it is. What I was getting at is this. Lets say the AI has 10 minitar kings and you have stacks of 1 MK and 100 MKs. The AI will go for the 1 MK. But what if the AI has 50 MKs, 75 MKs, 100 MKs? At what point will it decide to attack your 100 MKs? This could be easily tested and I might do it sometime. It also has to be considered there may be a random factor involved meaning there is some range where it could go either way.
Quote:
But! The last isn't true for auto-combat ai
What I think is interesting is that setting quick-combat in the preferences can turn out so differently than going into combat first and then hitting the auto-combat button.
Also I'm having trouble with Mr IK's "A" map. When I load saved game, I get a bunch of "could not find the sprite" errors. Examples of sprites are AVA0000.def, Zflame1, Zart142, Zobj007. This sounds like it could be a problem with game version. Properties on my exe file show version 4.0.0.0. (I have the US SOD version).
Lastly. I had 29 survivors in EX9. You thought you were being tricky with that one....but you can only fool me the first time
____________
|
|
IRh
Famous Hero
Lizard
|
|
posted November 27, 2003 06:08 PM |
|
|
Quote:
Sure the AI is pragmatic, but it could have a built in hate factor
No. I'm sure it IS. I think the AI uses ALL in its calculations of expected damage - hate (+50%, not 10% as I know), spells, skills... Except the random factor - er... - Or may be even this - AI in theory can KNOW, what would be the real damage! I dunno if they used it or just oriented by average value.
But, it was funny if hating crs made worse moves to kill their enemy... And if yours?
Quote:
At what point will it decide to attack your 100 MKs? This could be easily tested and I might do it sometime.
I think it is predictable, I could do some primitive math to discover that...
The idea is to predict the results of both behaviors and to compare the proportions.
Let I have 100+1 Angels, (As/AAs are the best for such computations - guina pigs) while AI has X. Att-Def are equal.
(AI thinks that: )
If he kills 1, we'll have 100 / X.
If he attacks the main stack (killing X / 4), retaliate kills... (100 - X/4) / 4 = 25 - X / 16. This makes (101-X/4) / (X-25+X/16)
(AI compares these fractions and makes decision.)
My guess: AI would strike main block if: 100 / X < (101 - X / 4) / (17/16 * X - 25)
Oh, too lazy to continue.
Trouble with A? So, It crashed? Sorry. It means it still needs WoG. It is SoD save, but was made from WoG. It runs properly on my SoD (3.1) as well as on WoG, cause I have WoG with these sprites (Z*.def). Strange, why are there references to these files?
Quote:
Lastly. I had 29 survivors in EX9. You thought you were being tricky with that one....but you can only fool me the first time
Er... I'll mark it 'Basic', then . Equal split?
|
|
binabik
Responsible
Legendary Hero
|
|
posted November 28, 2003 08:15 AM |
|
|
Quote:
(+50%, not 10% as I know)
Not sure you understood the +10% was in addition to the +50%. I wonder if the hate thing can be done in WOG. Or maybe make it like the berserkers in H4 where the player has no control. They go straight for the hated enemy....and if anyone gets in the way, give'em a whack, lose one hex movement and keep going.
Quote:
AI in theory can KNOW, what would be the real damage! I dunno if they used it or just oriented by average value
I suspect it would have to be one of those two cases, too many possibilites otherwise. For example 7 stacks of 1000 harpie hags each vs 7 other stacks. If you work it out, that's 21K possible outcomes for the 1st HH stack and approx 150K total for the round.
Quote:
....[all the math]....Oh, too lazy to continue.
You should have continued, then you would have found your errors But I still saw your point about proportioning.
Both the logic and math looked fine. But plugging in a couple values for X in the final inequality give negative values to the right hand side. Further values indicate either the sign is wrong ( > not < ) or something is REALLY wrong with the math/logic.
I won't admit how long I spent going over this and still not finding the errors.
Quote:
If he attacks the main stack (killing X / 4), retaliate kills... (100 - X/4) / 4 = 25 - X / 16
should read
"If he attacks the main stack (killing UP TO X / 4), retaliate kills... UP TO (100 - X/4) / 4 = 25 - X / 16
This takes care of negative values.
Quote:
AI would strike main block if
should read: "AI would strike SMALL block AS LONG AS"
Anyway, your formula gives a crossover point of 90-91. During play testing, the crossover was 83-84 consistantly with 10 tests each. The only thing I see in those numbers is at AI=83angels, AI damage=human player damage (20 dead each), at AI=84angels, AI damage > human player damage (21:19). That's still proportioning though. Your formula compares numbers left alive, this compares numbers lost. Too small a sample size to conclude anything though.
Quote:
Strange, why are there references to these files?
Speculation: If you install WOG it creates new executable files for WOG, leaving SOD executables alone. So you can play both (which I understand really is the case). But very likely the install can get by with replacing the SOD data files with WOG files. SOD would ignore any "unused" sprites in the data files the same as it ignores unused sprites in the original data files.
If you opened a WOG map with SOD it would be OK as long as only WOG sprites were used, but no scripts or other executables. When you save the map, there is no reason to strip away the sprites.
Quote:
Equal split?
Yep, almost two stacks left standing. I assume that's what you had in mind.
Now, speaking of math. Um...about ex4? I figured it out. That makes all 9 complete.
____________
|
|
IRh
Famous Hero
Lizard
|
|
posted November 28, 2003 03:44 PM |
|
|
Quote:
Not sure you understood the +10% was in addition to the +50%.
Ah, yes, I missed your point: you said AI should have considered the bonus 65% instead of 50%... But to make this effect noticeable, at least +200% should be used, IMHO. And, only for wandering armies.
And in WoG... Supreme ArchAngels hate Antichrists, done through the game code. But I don't know if it's possible for any creatures. At least, you can make a script that strikes additional damage after attack.
Quote:
For example 7 stacks of 1000 harpie hags each vs 7 other stacks. If you work it out, that's 21K possible outcomes...
7 Decisions at least and <~9M outcomes, with retal. But, you don'
t need these several M - what to do with them?
Really, AI counts 7 estimations and compares them. I doubt smth more than the average value (with some bonuses like killing the whole stack) can be used for the estimation.
AI compares killed numbers? Hmm, interesting observation. I thought comparing alive would be more logical... Hard question, what is better, really.
4 - well done ! Welcome to our ranks of Zealots ;P.
____________
|
|
binabik
Responsible
Legendary Hero
|
|
posted November 28, 2003 08:48 PM |
|
|
Quote:
AI compares killed numbers? Hmm, interesting observation. I thought comparing alive would be more logical... Hard question, what is better, really.
It's definitely inconclusive with such a limited test. But I have another piece of evidence I didn't state before, although still inconclusive.
With 83 angels the AI attacked single angle. I then gave AI magic arrow/advanced air and varied spell power.
The evidence indicates:
1) The AI's decision is based on whole units (partial damage rounded up).
2) The AI attacks main stack when (angels killed)>(AI angels killed on retal).
The tests showed evidence of some other AI behavior, but I won't go into it right now. I also want to do some more tests, again not right now. I'm preparing to leave town for 4 months....but I'll have my computer with me
____________
|
|
IRh
Famous Hero
Lizard
|
|
posted November 30, 2003 11:45 AM |
|
|
Curious facts.
Binabik, if you're still here, may be you have your own puzzles like mine ones?
____________
|
|
binabik
Responsible
Legendary Hero
|
|
posted December 01, 2003 05:42 AM |
|
|
Yea, I'm here. Just got checked into a motel. I'll be looking at these same four walls for 4 1/2 months!! (work related)
Turned out I don't have an access number for my ISP here. I'm using the EXPENSIVE way at the moment. Think I'm going to get a temp ISP, but will take a while to set it up.
Yea, I do have one map idea, don't know if it will be any good or not. I'll have to play test it.
Gotta run.
|
|
Client
Tavern Dweller
|
posted December 06, 2003 04:44 PM |
|
Edited By: Client on 6 Dec 2003
|
nebuka
100 master gremlins vs 100 sshooters (exercise 8)
Formation: 95,1,1,1,1,1,obsidian gargoyle
Round 1:
Teleport your first bait gremlin next to sshooters.
Gargoyle wait.
Sshooters kill the first bait gremlin.
95 gremlins attack.
Last 4 gremlinbites move forward, but make sure they move under the lake obstacle. (if you move them right forward the sshooters may get too close to your main stack)
Gargoyle defend.
Round2:
Teleport second bait gremlin next to sshooters.
Gargoyle wait.
Sshooters kill the second bait gremlin.
95 gremlins attack.
Last 3 gremlinbites move closer to sshooters.
Gargoyle moves as forward as possible.
Round3:
Mass bless.
Gargoyle moves next to sshooters as a bait.
Sshooters kill the Gargoyle.
95 gremlins attack.
Move all bait gremlins next to sshooters
Round4 and forward:
Now just keep resurrecting your baits and attack with your
95 gremlins all the time and bless them when they get unblessed. Never attack with your baits
btw this was a bit harder exercise than the other ranked basic tactics exercises
exercise 9 was easier than this imo..
can someone post something for the exercise 7.. can only kill up to 400 of them.. 
also exercise 4 and 6 seems to be impossible to me :<
____________
|
|
binabik
Responsible
Legendary Hero
|
|
posted December 07, 2003 02:36 AM |
|
|
Quote:
can someone post something for the exercise 7.. can only kill up to 400 of them..
also exercise 4 and 6 seems to be impossible to me :<
Are you SURE you want a hint? EX6 is easy. Look at your setup before going into battle. EX7 is not so easy. If you killed 400 it sounds like your basic tactics are probably right, but you're not using magic properly.
IRh: I'm up and running with a new ISP. I also installed WOG. I've only looked briefly at EX A though.
____________
|
|
Client
Tavern Dweller
|
|
posted December 07, 2003 12:18 PM |
|
Edited By: Client on 7 Dec 2003
|
omg I killed 1000 sshooters with 100, amazing 
used frenzy, bless, precicion and water elementals 
yes i would like a hint for exercise 6
____________
|
|
IRh
Famous Hero
Lizard
|
|
posted December 07, 2003 06:03 PM |
|
|
Quote:
omg I killed 1000 sshooters with 100, amazing
used frenzy, bless, precicion and water elementals
Srange... Without stone skin? Are you sure you have removed the bonus artefact?
Ex6: Frenzy... 1-Devil-stacks... Slow... 
____________
|
|
binabik
Responsible
Legendary Hero
|
|
posted December 07, 2003 06:26 PM |
|
|
Quote:
Srange... Without stone skin? Are you sure you have removed the bonus artefact?
I think I used stone skin at the very end with my last spells points. I don't know if it was necessary or not. I think the AI had around 150 SS against my 95. MY SS had bless and precision and first shot.
Quote:
Ex6: Frenzy... 1-Devil-stacks... Slow...
The only problem I saw with this one is if you have the main stack in a middle slot. The AA would move to the center of the battlefield first move. Even when slowed they can still attack every hex. With the main stack in slot 1, the AA would move further first move and leave room to attack/retreat.
IRh, I'm afraid I caused unneccessary confusion. When I said earlier that Mr Ik's "A" map doesn't work with SOD, it turned out I was using the "B" map. I hadn't even downloaded the A map yet. Anyway the A map works fine with my version of SOD. Sorry about the confusion, my fault.
With the A map, do I only need to take that rampart, or do I need to finish the whole map? I took the rampart with no losses, no berserk, etc. I also eliminated red with no losses. I stopped there, because this looks like a good map and I want to play it from the start.
____________
|
|
Client
Tavern Dweller
|
|
posted December 07, 2003 06:31 PM |
|
Edited By: Client on 7 Dec 2003
|
yup, without stone skin nor special artifact
I had one blocker 95 sshooters vs 146 with bless
Completed exercise6, the formation was important
the last one left is exercise 4, I have no clue what to do there..
____________
|
|
IRh
Famous Hero
Lizard
|
|
posted December 07, 2003 09:30 PM |
|
|
Yeah, 4 sems 2B the toughest...
And Stone skin - it increases effect of Frenzy.
about 6...
I used 1-creature stacks to hide my Devils at RIGHT corner of the battlefield.
Quote:
With the A map, do I only need to take that rampart, or do I need to finish the whole map?
Only to take rampart. I suspect Mr.Iks has dismissed some important heroes - may be they're necessary to finish a map...
I can send you a map if you mail me.
Quote:
Yea, I do have one map idea, don't know if it will be any good or not. I'll have to play test it.
OK. Would be good to see.
____________
|
| |
|
|
