Fall From Heaven unit power table

[Sareln]

I think the formula you've proposed could be a start, but as DaveV wrote it still isn't all that accurate. The priests definitely ought to have distinct power ratings, and Pyre Zombies are way more powerful than their base strength would suggest. Of course, stuff like mana available and whether a leader has cha/spi/arc/sum also affects the actual power of priests/mages. Would it even be possible to cross-reference traits when determining the power of a particular unit, or is unit power the same across civs/leaders?

iPower is Civilization and Leader invariant. It's an XML value that's referenced according to unitClass.

I could change the code to apply modifiers to a unit's iPower dynamically, but I was hoping to keep it static.

[Sareln]

iPower is Civilization and Leader invariant. It's an XML value that's referenced according to unitClass.

I could change the code to apply modifiers to a unit's iPower dynamically, but I was hoping to keep it static.

Dynamic Computation costs time while playing the game. If a calculation can be done ahead of time, that's when you should do it. That's part of why they compute the X/Y of the tech tree ahead of time, and it makes sense for static trees.

The AI does use the iPower numbers to decide on whether it feels weak or strong towards you. Think of iPower numbers based on how effectively an AI can use the unit. That's part of why the primary usage is strength.

Speaking of, I haven't given a way to handle defensive versus offensive strength... Ceiling of their Average?

[Cyneheard]

Speaking of, I haven't given a way to handle defensive versus offensive strength... Ceiling of their Average?

I think that makes sense. If you can incorporate the boost of civ traits into at least the UUs, that could improve the accuracy slightly.

I'd make the value of Channeling slightly higher if the unit is truly a caster (adept line, religious units) than if it's, say, a Fire Archer who only has one spell ever. I don't know the numbers well enough to have a good sense of what that means, though. Maybe +2/3/5 for All-Schools-Caster versus 1/2/4 for the Fire Archers and Vampires of the world?

And is doubling the strength based on 3 moves really reasonable? That seems slightly excessive. Maybe just a flat +1 (or +2) to power for each move above 1?

[DaveV]

Dynamic Computation costs time while playing the game. If a calculation can be done ahead of time, that's when you should do it. That's part of why they compute the X/Y of the tech tree ahead of time, and it makes sense for static trees.

But they must be computing an overall power each turn, and it couldn't be too much overhead to multiply by, say, 1.2 for connected copper and 1.4 for connected iron. A bronze warrior is essentially equal to a non-bronze axe/swordsman. My scale factors are lower than that comparison because some units won't benefit from bronze weapons, and because higher base strength units gain relatively less power from bronze weapons.

Also, you could multiply by 1.1 for aggressive (that free combat 1 really does make a difference), Ashen Veil (Ring of Flames), Order (Bless), Enchantment mana (enchanted blade), Death mana (level 1 summon!), Fire mana (Fireballs), etc.

[Sareln]

But they must be computing an overall power each turn, and it couldn't be too much overhead to multiply by, say, 1.2 for connected copper and 1.4 for connected iron. A bronze warrior is essentially equal to a non-bronze axe/swordsman. My scale factors are lower than that comparison because some units won't benefit from bronze weapons, and because higher base strength units gain relatively less power from bronze weapons.

Also, you could multiply by 1.1 for aggressive (that free combat 1 really does make a difference), Ashen Veil (Ring of Flames), Order (Bless), Enchantment mana (enchanted blade), Death mana (level 1 summon!), Fire mana (Fireballs), etc.

That's doable. I just want to avoid the unit-by-unit calculation each turn that this could easily slide towards - your computer would hate you for that...

[DaveV]

I just want to avoid the unit-by-unit calculation each turn that this could easily slide towards - your computer would hate you for that...

My computer is pretty weak, and I don't want to ask for something that leads to AI turns that are even longer than they already are.

[Square Leg]

Since there are so many factors in ffh that change the threat of a unit, I don't think it is possible to just look at strength. It is also probably not wise, for the sake of what Sareln said, to be doing unit checks at the start of each turn and having a fluid power system...

Call me old-fashioned but what is wrong with sitting a group of us down and going through each unit and assessing its power and manually assigning it? Outrageous no?

[xenin]

Some of those values are completely bizarre

UNIT_OGRE, 15
UNIT_CHAMPION, 18

UNIT_MESHABBER, 76
UNIT_MITHRIL_GOLEM, 100

I am wary of anything that will increase the computational cost of the power ratings - late game FFH on a large map is already brutal in terms of AI turns, so I think the key point is fixing the base values. A generic formula could work but it may be better to just do as Square Leg suggests and decide by committee. I have no idea how the underlying programming works for the power tables but intuitively recalculating power based on promotions seems like it will increase requirements. I agree with DaveV that it would still be useful (for units that can use those weapons) to get a power increase for metal connected though, as that seems like one empire-wide calculation and can make a big difference to a unit's actual strength. Scaling things by leader trait seems more effort than it's worth; I don't think it's unreasonable for people to know that their opponent is charismatic -> lots of promotions -> their troops are actually slightly stronger than the power graph suggests.

[DaveV]

I don't think it's unreasonable for people to know that their opponent is charismatic -> lots of promotions -> their troops are actually slightly stronger than the power graph suggests.

The problem is that the AI is neither a person nor reasonable, and I think it bases a lot of its decisions on power ratings.

[xenin]

The problem is that the AI is neither a person nor reasonable, and I think it bases a lot of its decisions on power ratings.

Sure, and if there's an easy way to code charismatic leader = 20% more power or whatever then that's definitely an improvement over the current system; I just feel an imperfect measure beats a perfect one that slows down the game.