Indeed. I had my butt handed to me in a game last week on hard difficulty and I immediately set it back to normal difficulty after that.
The AIs are relentless. They just keep coming. In that game, I was hopping my best 3 stacks from town to town trying to keep them from taking them and I held back the tide for a while and killed zillions of their good units but it was hopeless. They got too far ahead of me in population and production and went for the throat. They had a lot of high level units which really made it tough.
This really is the best AI I’ve seen a a long, long while in a strategy game.
Yes! I always stick +accuracy equipment on new units.
It’s an interesting decision whether or not to materialize your deity. Not having mine materialized saved my bacon today. I had my main army far away from my empire taking out an AI when another AI showed up. A combination of my deity’s fireballs, towers and archers easily held them off. If they had their deity though it would have been a different story. The AI was real smart in avoiding some of my more defended cities, it goes for your weaknesses.
I am already thinking about my next game. I’m thinking of starting with a lot of spell books in a single path to get more powerful magic sooner. I went 4 3 this game, combine that with not knowing what I was doing and my tech was limited. I am still not at any summons.
My strategy these days is to start off etherealized and support my main stack while they look for items and spellbooks in lairs/dungeons. I like starting that way because it costs 5 deity ability points to start materialized, which is a waste of starting points, imo.
Once I have some basic offensive and cc spells researched and enough meditation and battle mana points built up, my B unit stack is about ready to roll out and my A stack is strong and has magic user units that can cast spells on their own. So A stack deity support is not really needed anymore.
At that point I materialize and use my deity to level the B stack and himself up to around level 5. At any point if I need to defend my cities or my A stack, I can just cast etherealize (which is instant) and on the same turn I can defend or whatever. Or, I may also decide to etherealize again if I find some new spellbooks and want to research the juicy spells inside. Then once I do that I materialize again. So, I bounce back and forth between the two states a lot after the first 100 turns or so depending on what I need my deity to do at the moment.
About the starting number of spellbooks: I never bother to research the spellbooks themselves, just the spells in them. It’s so much faster. For that reason I only start with 2 spellbooks. One of each spell school I want to start with. Then I pick the ‘meditator’ and the ‘battle caster’ deity abilities. They give a big boost to the speed of increasing meditation and battle mana. I can always find the spell books in-game but I can only get those sweet boosts at startup. As an example, I started with 1 fire and 1 holy spellbook in my current game. Now, at turn 385, I have:
21 Holy
11 Alteration
9 Energy
8 Fire
6 Poison
3 Arcane
3 Cold
2 Nature
So, I’ve found all but 2 of those books in dungeons. I spend so much time in dungeons looking for old books that I’m a regular Indiana Jones.
I used my fire books to summon a baby red dragon early on and put him in my A stack. He evolved into a young red dragon which are great because they get the fireball spell at level 5. He’s 5 levels away now from evolving into an adult red dragon which will be pretty fun. From the Holy spellbooks I learned summon angel and have one of those in my main stack too. I takes forever to evolve them to archangels but I just found enough Holy spellbooks to get the summon archangel spell so I’m currently researching that.
Archangels are great because they can summon angels. And angels can summon lesser angels. And all of them can heal.
To me this sounds like AoW and Dominions had a baby?
(Aka the game I have been dreaming off since I fired up Dominions and thought how cool it would be to have your player character in an AoW game be an actual Dragon, or lump of rock, or Giant Monkey…)
Not as ugly as Dominions, not as pretty as AoW, a different but equally good combat system?
An opponent who has more population, a larger economy, a larger military etc relentlessly grounding you down isn’t good AI IMHO.
Fire up Aow3, go on an extra large map, play the game normally, typically you’ll take out 2 or 3 AI before the remaining ones are so much larger than you, and have so many armies, that the game becomes a bit of a slog and you eventually die, because you are fighting one or 2 epic fights every turn, for easily a dozen plus turns.
That said, the fact that the AI here at least recognizes it’s numerical advantage and comes after you is not a bad thing at all.
All the more if it works consistently along all game settings, because it certainly appears there are plenty of ways to set a game up.
It can see into fog of war and also gets resource advantages depending on the difficulty setting, according to the developer. They also have extra AI code and settings for tougher AI at higher difficulties.
The AI ignores fog of war on the world map, so it will always track you when you are within its “search” range (which is quite a large number).
The AI Difficulty is mostly a resource cheat. It basically works as though the AI were playing at a different game speed (more gold, mana, production, research).
And there is some enabled code to make the AI more evasive in battles on harder difficulties.
There is also a game setting to make the AI more “intelligent” in battles on harder AI difficulty. They said they had disabled it because:
“… it really didn’t play out as that fun to have the AI hard target your weakest units and kill them off 1 by 1, so we disabled it.”
However, folk said no, we really want that back in there so they added it back. It’s just turned down by 50% in the default config settings file. Folks can tweak it up if they want more challenge.
So, the bottom line is that the AI gets advantages, which is no surprise. However, they seem to have done a really nice job coding it.
I assume the AI gets extra resources, my early impression is that the AI “knows” what to do with those resources. Though I am a noob to this game so I am far from optimizing play.
You can mostly just let it do it’s thing, especially at the beginning of the game. However there are a few items of note that you’re going to want to research at some point:
Road Building - This increases the maximum number of roads you can have in a hex. The more you have in a city hex, the more workers you can put there. It will really increase the output of the city improvements you build in those hexes. Researching this has the added bonus of leading you in the research tree right to World Travel 1 which increases the world movement of all your units by one. Very handy.
Farming Techniques - This adds levels to your farms. So if you research this up to Farming Techniques 10, you can build 10 level farms, which put out a TON of food. Great for growing your pop to high levels. But you won’t need this at the beginning of the game.
Note that if you want to build a transporter device in your cities, you’ll need to first research Transporter Device Knowledge in civics and then you can unlock it in the city building tree.
The rest of the options are more situational and on an as needed basis. For example, if you need more mana and you have a lot of piety in your cities you might start researching Mana Link to increase it. However, for the most part I just let it do it’s thing unless i feel the need to step in and prioritize something else.
Regarding the AI, a while back one of the devs posted how the piece of the AI that targets your units in battle works. I found it pretty interesting.
Currently an AI unit does the following (roughly) each attack to determine its target:
Checks the expected damage it will do, with each of its attacks, against every enemy unit on the battle field. It takes into account armor, resistances etc. Call this number the ai_target_weight.
It modifies the ai_target_weight by certain abilities that it has, and certain abilities the enemy has.
For example, enemies with mana regeneration have their ai_target_weight increased a bit, as they are more threatening. While flying enemies that cannot be hit have their ai_target_weight reduced to almost 0. While units that normally fly and could not be hit, but are currently immobilized, have their ai_target_weight increased drastically to take advantage of their short-term vulnerability. Passive enemy units have their ai_target_weight reduced to almost 0 (which is why workers and settlers are almost always attacked last). Enemies behind walls that are too hard to hit won’t be targetted when ammo gets too low. Units with Fire immunity will never be targetted by attack which do fire damage ie the ai_target_weight for this enemy/attack combo will be 0. Sleeping enemies are less likely to be targetted since attacking them would awaken them, etc. There are a huge number of modifications that occur here.
Now every ai_target_weight is reduced by a certain percentage, with “dumber” units reducing the ai_target_weight amount more than “intelligent” units. This results in dumber units having a tendency to more equally weight enemy units and be less likely (relative to intelligent units) to target the enemy unit it will do the most damage against. So if a heavily armored enemy and a squishy mage enemy are standing next to each other, an intelligent unit is more likely to target the squishy mage with a melee attack than a dumber unit is, and vice versa (a dumb unit will often target a heavily armored unit rather than a squishy mage). The settings ai_battle_target_weakest_enemy_chance in the game_settings.txt also affects this. Notice that ai_battle_target_weakest_enemy_chance doesn’t refer to weakest in terms of lowest health, but rather weakest in terms of “squishiest” enemy (and may have been better named ai_battle_target_squishiest_enemy_chance).
Next the ai_target_weight is reduced proportional to how many turns it would take to get into attack range of the enemy unit. So if it would attack take 6 turns to get to the squishy mage, but 1 turn to get to the not-as-squishy lightly armored archer, it will be much more likely to target the archer. This is because it would be able to get 5? attacks on the archer before it could even get 1 attack on the squishy mage. And it will also be getting hit for all these turns from the archer and possibly the mage, while it won’t be doing any damage to either for these turns. This is why closer units are most often targetted over further units.
Lastly the ai_target_weight is randomized a slight bit (up or down by as much as 20%?), to add some unpredictability.
Now the AI unit will choose its target and its attack based on the ai_target_weights of every one of its attacks/enemy unit combinations. It will lock onto that target for ai_battle_lock_on_target_duration combat rounds (so it won’t go through all the above calculations again until its target dies/disappears(invis) or some other reason).
The AI unit will move towards its target. Once its moves are done for the combat round, if it cannot attack its designated target, then it will reevaluate every enemy in range that it CAN attack immediately, and MAY attack one of them (but it may not; for example it won’t shoot if it is low on ammo and the chance to hit an enemy has been reduced too much by moving, or because the enemy is behind a wall… lots of stuff is done here as well).
That’s a basic summary of how the AI currently targets an enemy unit. Although I’m sure I’ve forgotten a few other important things the AI does.
Resource bonuses are Ok and expected, but I’m not a fan of the AI seeing into the FoW. It makes it impossible to be sneaky. I also don’t like when an AI makes a beeline for some city or stack that it shouldn’t know about. I guess if they are clever in how they implement it so it doesn’t always happen and it isn’t obvious then it can be OK. Age of Wonders AI can see unit stacks it shouldn’t and it is still a great game.
I may get this on sale since much of what you guys say here make it sound interesting.
Back on topic, I look at screenshots on steam. Game looks not very good. Does this get in the way? By way of comparison, Dominions isn’t a looker but it was never an issue.
Not for me. I do wish the graphics had some of the charm of Master of Magic. At least it doesn’t go for cutesy, bobble heads which is where I draw the line.
CS, nice strategy. I never thought of going back and forth to materialized. I figured there’d be a cost/cool down. I need to work on my dungeon game. I just tended to go into easier dungeons, auto-battle and leave after the first level. Not knowing what I was doing and having weak troops and magic played into this.
BBB, comparing this to a cross between Dominions and AoW is spot on. The combat system is turn based like AoW3.
I agree with CS on civics research. I’ve been going for the research and production bonuses.
As for the graphics, that’s what kept me away at first, but after actually playing it, I’m OK with them. There are already some mods that improve them a bit.
Going back to etherealized is instant and costs nothing. Materializing time depends on how much meditation you have. That’s why I don’t do this until I build it up. It takes a turn or two then, so not bad at all.
It’s the same as with Dominions, it’s not an issue. It grew on me after I started playing. The unit art is actually better than Dominions.
That looks worse than Dominions, not better :P. That bear, that beholder ugh.
It looks amateurish for me. Literally. I had some friends that with 14-15 years drew at that level.
