Any improvements in the tactical part of the game?

[Twan]

I'm a little disappointed to find nothing about this in the announced features, but for me the 1st priority for an improvement of dominions is the tactical part of the game : how combats work, especially after round 5. If the unpredictable results have some charm, I think that long battles are too much out of control in dom2, making victory too often the result of random choices made by the tactical AI, when most important fights are 10+ rounds long.

Is the scripting for mages going to continue to be limited to 5 orders without more precise options than "cast spells", "attack" or "retreat" after ? (I hope there will be a final order like "cast summoning/dammage/buff spells in priority", or a "repeat orders from x" option -seems easy to implement and more precise, after the 5 first rounds the mage repeat for example the orders 3,4,5 as long as he can-). And/or consigns like "never cast spells of this school/path" (to say for example to a water/fire mage to use only his fire spells as he's with creatures sensible to cold), "never use gems after your script", etc...

Or will the tactical AI for mages be improved at least (or are we going to see mages continue to cast all the shields they can after round 5 even when the closest ennemy is at 40' and engaged, or to kill their own companions by casting heath or cold auras, or to summon lots of useless troops when they should use damage spells, as the troops are blocked at the entrance of a fortress ? or to make strange uses of their gems, etc...) ?

Will the orders for the troops / commanders remain all the same (without an option for example to say to a pack of mounted archers to fire and use their superior movement to stay at range -the first time I played dom2 I was thinking that "fire and flee" was for this kind of guerilla, and was very disapointed to see my skirmishers definitively retreating instead of fleeing melee to continue to harass their opponents- , or to say to a mage to go at the appropriate range then cast his spell -how many short range spells are never used in scripts because of the lack of this order ?-, etc...)

Anyway, I will buy and play dom3 even if there is no improvement in this area, and I'm pleased by all other announced features and to see such a lot of content added to the game, but I will be really decieved if the battle system / scripting / tactical AI are all exactly the same (what will mean even more chaotic battles in dom3, as we will have more troops the fights will be longer, but if it's with the same options / scripts limited to 5 rounds it's the out of control phase that will be longer).

[Chazar]

I agree.

However, while it seems to me that sixth-stage commands like "cast summoning/damage/buff" or "cast damage/buff/summon" should be easy to implement or add (since each would merely require an additional set of number costant definitions for the tactical AI spell selection, i.e. a different weight for each spell), I doubt that we will see such improvements:

1. While it would be nice to have this control, it certainly means longer turn times for human players: In Dom2 this tactical control is already important, but it could be neglected in many cases with a player using some template scripts saved to the scripting-shortcut-keys. This speeds things up. A powerful tactical setup means that one really has to set up the army again and again for each battle individually. This is clearly a good and a bad thing... http://forum.shrapnelgames.com/images/smilies/tongue.gif (I would like it, but I would never finish a turn anymore since I suffer from being pedantic... http://forum.shrapnelgames.com/images/smilies/wink.gif )

2. Such control will most certainly exploit much more inbalances within the game. A game with thousands of monsters and hundreds of spells just cannot be completely balanced without being boring. More tactical control will allow players to exploit this, making balancing even more difficult. The tactical AI evens things out, because everyone has to live with it. It helps the game to appear well balanced.

3. More tactical control will make playing against computer controlled nations probably more easier. In my humble opinion Dominions is and should be a multiplayer-only game. However, the single-player community seems strong from what we can learn on this forum, so while everyone might welcome more choice for the tactical AI, players will then complain again about the "bad AI", for whatever that really means. For me it's good enough to learn the game or to test a map...

So, summing up, for these three reasons, I do not expect any major improvements for Dom3 in this aspect of the game. Maybe some commands to make light cavalry units useful somehow, since this has been extensively criticised for Dom2 in past time in my perception. But who am I to guess? Let us wait and see...

[Oversway]

Having even minor improvements in the battle scripting would be fantastic.

[NTJedi]

It would be nice even if the scripting went from 5 commands to 7 commands... I look forward to seeing what the game has available.

[Gandalf Parker]

I dont think that a features list has been posted for Dom3.
It might not even be finalized.

I have heard the devs make references to AI so apparently some attention is being put there but we wont get to debate any of it until the release. We might want to return to the Dom2 forums and develop some real AI strategies, then test them. That might be helpful.

[Argitoth]

The battles in Dom2 make up at least half the game, if not more. It should have been on the top of their "To Improve" list before they began Dom3. I would be very surprised if we didn't see some kind of improvement.

[Graeme Dice]

[quoteor to summon lots of useless troops when they should use damage spells, as the troops are blocked at the entrance of a fortress

[/quote]

So how, specifically, would you test for this particular case without screwing up the other cases where you would want summoning to go on indefinitely?

[Twan]

I've listed several possible improvements but of course in some cases it's one of the other.

If there is a "cast summoning spells" combat option after the 5 scripted orders, it's up to the player to use or not this order. If he's storming a fortress with a sufficient number of troops his last order should be "cast dammage spells" or "cast buff spells".

If the AI is improved, it's one the hundred situations formulaes should be made for. Don't know how the tactical AI works in dom2, but I think the choice of a spell should be based some kind of situationnal value, and of course a storm fort situation is not the same as a normal fight in open lands.


Something like :

value of a summoning spell = [(value of ennemy troops + value of summonable unit - value of your already engaged troops) with a minimum, for example value of ennemy troops /10] x situationnal number for the summonable unit type

("value of troops" = (hp + medium stat) x number of units for example or other quality and quantity based formula)

And the situationnal value of melee troops when storming fortress should be something like (size of the gates/number of already engaged melee troop) + (0,1 x number of losses last turn), when the situationnal value of flyers is for example 1.

In this example if you have already 3 x more melee troops than the number the size of the gates allow to fight and not lost any in the last round, the value of a spell summoning melee troops will be only 1/3 of the value of a spell summoning flyers, and probably far inferior to the value of a dammage spell based on another formula (taking in account for example the number of ennemies at spell range, their vulnerability to this spell, etc).

[Gandalf Parker]

It sounds good. At least its 200% better than the ones who say "broke" then when asked for suggestions have nothing more than "make it not broke". http://forum.shrapnelgames.com/images/smilies/happy.gif
Have you tested it? I cant tell from looking at it if Id want Johan to spend a ton of time programming it in then have the beta-testers spend a ton of time testing it.

Try putting it down on paper, even if you have to guess at some variables. Then try playing a game with you, and one other nation which you swear to play only by the thing you wrote down. And remember to try it with most of the nations (it does no good to come up with a formula that makes Ulm stronger and Caelum weaker). Also keep in mind small-map tactics dont always translate to big-map taces and vice versa.

If we could get working formulas for summoning, troop purchases, mage purchases, research, and structure building then this could be a game with decent solo-play along with great multi-play. Of course it might make the hosting take a couple of hours on big games but I could actually play with that.

Id be interested in the results.

[alexti]

How would he test it? AFAIK, there's no way to control commander's spellcasting beyond turn 5. Neither there's a way to query the situation in the battle. I'm not sure about the general suggested formula, because in many cases it won't be particularly good plan, but curbing summoning in easier cases, like storming the castle, with this approach should work.

[Gandalf Parker]

True. Unlike most of the formulas this would have to be more subjective. Watching battles and then trying to decide if the formula would have made a difference in that specific battle.

Abit more difficult in this case. But the "test it" method has helped alot in the past to show us times when a fomula on paper didnt tend to work out that well if you tried to robotically follow it for all nations

[Twan]

I think the best way to make such tests possible would be to release a round per round dom3 combat engine (functionning like the heroes serie or MoM tactical combat, you choose the spells casted during the fight) as a minigame. It may be very usefull for testing AI scripts as well as a good addition to the game (I imagine that many players would be happy to make duels on this mini game, and it may even be made a game option to use this round per round engine for normal combat resolution -of course only in non pbem games-). The other possibility being of course a scriptable AI you can test with the normal combat engine. But I know the chances to see one of the other thing released in an easy to use by players form are very small considered the lot of work needed.

Anyway making working formulas without the hability to test them is hard, so I can't propose anything precise, but I may list some factors I hope to be considered for the value of each spell when the AI have to make a choice.

General :
- the fatigue level of the mage (any spell making the mage reach more than x fatigue should be discarded first). If the game allow more player choices I hope this x to be based on a player choice (not for each mage as it's too much microgestion, but why not having general consign for all our mages like "rest if your are tired" -your mages will hold if their fatigue is 80 or more-, "stay active as long as possible" -the mages won't cast any spell making them reach 100 fatigue, except if they can't cast anything else-, "be prudent in spellcasting" -the mages won't cast any spell making them reach 130 fatigue-, "use your powers" -the mages will use any spell they juge usefull without fatigue limit-.
- the gem use compared to the number of gems the mage has (as well giving the player the hability to give a consign for his mages about gem use will be appreciated)

Summoning spells : your number of troops and their relative power compared to the ennemy number of troops and their relative power (the more you are outnumered the more a summoning spell is valuable) the losses of the last round ; in storm castle situations the size of the gates for the spells summoning melee troops.

Dammage spells : the number/relative power of ennemy troop you may affect at best (at good range / in the ideal area of effect -if your precision roll is successfull-), their natural resistance against this kind of spell (but not their resistance improved by spells the ennemy had casted -or it's too much an advantage for offensive casters).

Resistance/defensive buff spells : the nature of the spells or attacks used against your troops in the last round, the dammage taken by attack/spell type * the number of troops the spell may improve at best (ie if the troops with an earth mage had suffered 40 melee dammage and 80 mind burn dammage last round, the iron will spell will have a bigger value than iron warriors, but legion of steel or marble warriors will be preferred as they affects far more troops)

Offensive buff spells (strength, flaming arrows, weapons of sharpness etc) : number of affected troops, quality of the improvement, may be determined by the dammage your troops may have done with this spell if active last round (ie if last round with "weapons of sharpness" your units would have made 50% more dammage and with "strength of giant" 20% more, "weapons of sharpness" will be chosen in priority).

Heal spell : number of troops affected, the medium dammage they have taken, the % of their hp the spell will give back * the offensive power of these troops (it's more interesting to give back 20% of his hp to a spring hawk than to give back 3% to a vine ogre).

Personnal melee preparation/shield spells : the proximity of the closest ennemy and his speed, the presence of flyers and cavalry in the ennemy army, the number of ennemy troops compared to yours (if you have more troops and the ennemies are far away and with no flyer nor cavalry, your mages have no reason to use "astral weapon" or other "fire shield").

Of course the big difficulty is to have all these formulas based on these very different factors giving final numbers in the same range, comparable by the tactical AI to make a choice, this game needs a genious biclassed mathematician /AI programmer to really reach all its potential in the tactical area.

[Dhaeron]

I don't think the AI can be programmed to actually make clever choices. I think for every scenario where it makes the right choice based on such rules there's another (or more) scenario where it totally screws up because of that.

One thing i'd really like to see that'd kinda fix this problem is the possibility to script ongoing order for troops. I.e.: cast (specific soell) ad infinitum.
Or maybe less specific "only cast support/damage/summon/protection spells".