Sleep and zombies, or zombies in general

[Seravy]

For each unit with create undead or life steal, add points equal to its cost (the one used to determine experience and AI stack building rules, since there are three different costs used in the game). If the wizard has Death books, add Skill*Death Book count/5 to it. Mark units on a first come-first serve basis until this total cost is reached, to be raised. (if that unit is immune to being raised, it'll be still marked and wastes the points)

[Seravy]

So when the AI does it against another AI, the human sees that and says 'that is a cool, appealing result! I want to do that too!'

Yeah but that's the problem, they aren't doing the same thing. It looks like it's the same thing, and if it was a PVP game it would be the same thing, but it's a PVE game so it isn't the same thing.
The AI is taking them away from the enemy of their enemies = their potential friends. The human is taking them away from their enemies. Huge difference.

Flavor-wise it's the same thing. But the net gameplay result is a sum of zero for the AI, and a massive net gain for the human. Which aren't the same. So they obviously can't have the same rules because game rules are based on the GAMEPLAY. Otherwise you get a flavorful, but unplayable game.

(and yes, the human having an ally and another AI stealing units from that AI can be a thing, but it's extremely rare...)

...not to mention the difference in potential the same high tier unit represents in the hand of a human, vs an AI.

[Nelphine]

Jeepers, yeah the AI is getting like 9 times the amount of undead I've been advocating for. That certainly wouldn't work for the human, but it also implies, the AI may actually be getting too much, which is one reason why I actually see this as a problem.

Also, AI has no difference between create undead and life steal, when for the human it's a major difference. However, given that the AI can't plan to do things like attack a unit only with ghouls while the non ghouls attack a different unit, that means in tactical the ai are much worse than the human right now. In theory this is a wash, but it means the AI loses out in early game and gains only in the later game during strategic combat (which is much rarer in late game), this means the current ai kind of have the worst of both worlds - they're weak in early tactical when ghouls and zombies have the most impact, but they get extra in strategic late game when the human has lost their own ability - however, since strategic is so rare then, they aren't actually getting a lot.

The fact that strategic has to look at books not spells exacerbates this, but as long as the difference between life steal and create undead isn't addressed, the books vs spells doesn't really matter.


However, I think the books vs spell thing is actually fine.. except, multiple books shouldn't impact this,extra books doesn't have anything to do with how many undead you can create.



So, overall, strategic is still creating too much which is part of the problem (notably, once rare is reached, the lack of difference between create undead and life steal becomes relevant AND casting skill has reached big enough at that stage that the arbitrary multiplier for books makes it very relevant.)

However this is somewhat balanced by the fact that the AI doesn't plan to do any of this, and so often the AI will mix units that dondont create undead with units that do. But again, at rare, wraiths only stack with themselves (and angels), so this emphasizes the difference - the ai, for strategic purposes has the perfect undead creation stack.


While I understand converting units instead of killing them keeps the game more dangerous, strategic combat damage calculations are already designed to make the winner more dangerous during an AI war. Adding this level of strategic undead creation effectively doubles up on that, making a late game ai vs AI war, where the winning side is death, FAR more dangerous to the human than any other AI vs AI war.


So.
Conclusion.
Part of my problem comes from the ai simply creating far too much undead late game. I do want that, but looking at it, no, it wouldn't be balanced - and I don't think the strategic combat formula is balanced.

I'd suggest making life steal different from create undead, and make life steal units only add half their value. The only reason for this is because the ai is shafted by early tactical undead creation, so they need that balance from strategic, otherwise I'd just half all of it. Conveniently though create undead is early and life steal is late, so just halving life steal has the correct impact.

I'd further half the impact of casting ability. (Make it div 10 instead of div 5).



Then I'd use exactly the same for tactical, except I'd give it an extra divide by 2 modifier.


So tactical and strategic would work exactly the same way, but tactical would create half the undead. (If you could make a picklist to let the human choose what to raise instead of first come first serve, I'd also do that.)

And yes, books shouldn't actually be the basis. Since you already track it, for tactical , instead of books, count how much damage is done by life steal abilities, and use a conversion of life steal damage to value, and replace the book/casting will modifier with the life steal damage/value in the formula. (So, for instance, if you have 20 skill, and cast 2 life drains, we would expect that to be ~6-7 damage against standard targets, so, use a modifier between 3 and 5, so 2 life drains doing 6-7 damage would add 18-35 to the value of how many undead were created.)

[Bahgtru]

(this seems to have been missed)

Fortunately all the spells involved are commons - if you are actually playing heavy death where undead creation is the theme, you can't possibly miss them. Playing 2-3 books is a different story but that doesn't deserve the additional chance anyway. At 6 (or 7?) books you have every common, and even if you have fewer books (which you can't really do playing mono-Death due to the 4 retort limit), as long as you are mono-death, the common spells in treasure will automatically fill the holes, even if you never trade - they can't be anything else but those 2-3 spells you missed. Of course if you play multiple realms, that's a different story, then you have to pick the spells, but that's the price for playing more realms, you are weaker in each of them individually.

No that's not the point - 2-3 books don't rely on having 2 specific commons (or choose them at the beginning), so if this is meant to nerf multi-book setups it doesn't work. (I'd suggest a dedicated discussion if you think that's wanted btw)

With this choice you remove options to mono-death mostly - multi-realms have other ways like the ones listed before. For example:
- High armor or HP, resistance 4-6 city troops (examples: beastmen, trolls, pikemen, phalanx, etc): you make it much more difficult for mono-death to obtain these as undead, and in the mid game this can be a big set of your battles. That translates into a slow-down of your strategy.
- flying creatures: sleep is a way to deal with them. So there are cases where the change makes it obligatory to choose between dealing with flyers and getting undead, and life drain can't cope with all of them. Flying is already OP enough...
- Floating island fights?
I could probably come up with more, but the big deal is the first one. All in all, even a randomised undead generation as suggested by Nelphine under sleep would be better than losing this option. (incidentally, it could be made into the same mechanic as werewolves). To his idea you answered that randomness is rage inducing... But how is the randomness of the undead generation any less rage inducing than that of the save or die spell itself? Losing a hero to sleep is much more rage inducing, and those don't get raised

Why don't we try to think about the improvement to auto at least? Write down your thoughts on the procedure as it is now and let's give it some thought before deciding. That idea has gained support by others in this thread, and I know that you like to teach the AI tricks.

[Seravy]

However, I think the books vs spell thing is actually fine.. except, multiple books shouldn't impact this,extra books doesn't have anything to do with how many undead you can create.

It has. Assuming the strategic combat tries to at least simulate the real combat, a wizard with 2 Death and 8 Chaos books will cast death spells much less often and chaos spells much more, resulting in fewer undead, while a wizard with 8 Death and 2 Chaos will use mostly Death spells and minimal chaos instead.
Obviously it's not that simple but it's the closest approximation we can get - more books do increase the likelihood of casting spells from that realm.
(albeit yes, there are no actual rare spells in the game that are in-combat and make undead, so this isn't all that good as an approximation, still, better than nothing - the "more nondeath books = fewer death spells" part still holds.)

Conveniently though create undead is early and life steal is late, so just halving life steal has the correct impact.

Agreed. I see space issues but I might be able to cram this much detail in somehow.
...or not, I mean looking at the actual cost of units affected, it doesn't seem necessary.
Ghouls are 100, Wraiths are 360 - but a wraith is indeed capable of making 3.6 times more undead. (more like, you'd need 4 ghouls to convert the one thing the wraith does alone)
It's only really off for Death Knights and Demon Lords which cost more than a Wraith but aren't better at making undead. But by the time those come into play, it's kinda unlikely to for the ability to be relevant - lairs are mostly gone, and it's probably only 1 AI still standing.
I don't really have space to add so much detail as to make the very rares different from Wraiths so it's best if I don't halve it.
If the ghouls vs wraiths value ratio was a multiplier of 6 or greater then I'd see the change justified.

I think the ratio of create undead vs life steal is about right. (zombies are worth 70 btw)
Of course, halving both, if we think it's too much, is an option.

(Vampiric weapons on heroes is also a thing, but then the converting capacity entirely depends on what hero has the item. So it can be anywhere from 100 to 999.)

I'd further half the impact of casting ability. (Make it div 10 instead of div 5).

Not sure about that one. At 10 books and 100 skill the current result is 200 value - you get to convert an uncommon. 100 skill casts Syphon Life 3 times, or Life Drain 10 times, so that is realistic.
Reducing it to 100 (no uncommons, only 1 common or 2 if they are really weak) would be harsh. Even most decent normal units are over 100 cost.

count how much damage is done by life steal abilities, and use a conversion of life steal damage to value,

But that's bullshit. Measuring damage against unit cost is stupid. No sane player would ever accept that, myself included. What if the unit has too low hp to fully cover their cost with damage? Then it can't be raised at all. What if it's low cost and easy to damage? Then the extra damage carries over and raises more stuff. I see potential for abuse here as well - a regenerating lizard in an army filled with archangels. Keep syphoning the low resistance lizard to gather damage "points", thanks to regeneration I can drain up to 80 from it. Then that translates into 3 undead archangels for me why?

Admit defeat, these ideas get crazier every time.

[Seravy]

(this seems to have been missed)

Fortunately all the spells involved are commons - if you are actually playing heavy death where undead creation is the theme, you can't possibly miss them. Playing 2-3 books is a different story but that doesn't deserve the additional chance anyway. At 6 (or 7?) books you have every common, and even if you have fewer books (which you can't really do playing mono-Death due to the 4 retort limit), as long as you are mono-death, the common spells in treasure will automatically fill the holes, even if you never trade - they can't be anything else but those 2-3 spells you missed. Of course if you play multiple realms, that's a different story, then you have to pick the spells, but that's the price for playing more realms, you are weaker in each of them individually.

No that's not the point - 2-3 books don't rely on having 2 specific commons (or choose them at the beginning), so if this is meant to nerf multi-book setups it doesn't work. (I'd suggest a dedicated discussion if you think that's wanted btw)

With this choice you remove options to mono-death mostly - multi-realms have other ways like the ones listed before. For example:
- High armor or HP, resistance 4-6 city troops (examples: beastmen, trolls, pikemen, phalanx, etc): you make it much more difficult for mono-death to obtain these as undead, and in the mid game this can be a big set of your battles. That translates into a slow-down of your strategy.
- flying creatures: sleep is a way to deal with them. So there are cases where the change makes it obligatory to choose between dealing with flyers and getting undead, and life drain can't cope with all of them. Flying is already OP enough...
- Floating island fights?
I could probably come up with more, but the big deal is the first one. All in all, even a randomised undead generation as suggested by Nelphine under sleep would be better than losing this option. (incidentally, it could be made into the same mechanic as werewolves). To his idea you answered that randomness is rage inducing... But how is the randomness of the undead generation any less rage inducing than that of the save or die spell itself? Losing a hero to sleep is much more rage inducing, and those don't get raised

Why don't we try to think about the improvement to auto at least? Write down your thoughts on the procedure as it is now and let's give it some thought before deciding. That idea has gained support by others in this thread, and I know that you like to teach the AI tricks.

No, it wasn't missed, I was just really tired of this. At this point this discussion has probably taken up more of my time than the entire implementation of the building queues system which was supposed to be mythical tier difficult.

Units with more hp should be harder to convert, that makes perfect sense - they are better units most of the time.
Flying, I see no problem there. While you're losing, they will attack your units, zombies included. When you're winning, you can afford using the Life Steal and Syphon Life to finish them off. Yes, it's inconvenient but see the previous case - flying units are better so there is nothing wrong with converting them becoming harder.
No idea what you mean by floating island fights - summon zombie is a nonsea summon, you can't cast it there.
All this proves is we are doing the correct thing by not allowing the player to completely ignore the stats and abilities on the things they want to convert anymore, and since Sleep did that, it was broken and unhealthy for the game. 

If we do want "resist or turn into undead" as a spell, we can design one instead of trying to fix a poorly working combo that has all sorts of issues and abuse potential. But we don't want one - it would be redundant and less ideal than the existing "save or take create undead type damage" spells.

[Nelphine]

It has. Assuming the strategic combat tries to at least simulate the real combat, a wizard with 2 Death and 8 Chaos books will cast death spells much less often and chaos spells much more, resulting in fewer undead, while a wizard with 8 Death and 2 Chaos will use mostly Death spells and minimal chaos instead.
Obviously it's not that simple but it's the closest approximation we can get - more books do increase the likelihood of casting spells from that realm.
(albeit yes, there are no actual rare spells in the game that are in-combat and make undead, so this isn't all that good as an approximation, still, better than nothing - the "more nondeath books = fewer death spells" part still holds.)

No, more death books don't increase how much of your casting skill goes into death; instead, non death books reduce how much of your casting skill goes into death. A person who has 1 death book, or 10 books, but is mono death, still spends all of their casting skill on death.

So what it should be is [casting skill]*[number of realms the wizard knows that are death]/[number of realms that the wizard knows] and then divide by some number to account for roughly how many of the death spells are actually life stealing spells.

For instance, under the current formula, wizard A has 4 death books and 4 chaos books and 200 casting skill. A then gets 200*4/5 = 160 added value to undead creation. Wizard B has 4 death books and no other books. B then gets 200*4/5 = 160 added value to undead creation.

Under my formula, wizard A would have 200*1/2 = 100 base casting skill assigned to death spells; then we divide by some number to account for roughly how much of the death spells are actually life stealing spells. Wizard B would have 200*1/1 = 200 base casting skill assigned to death spells, then we divide by some number to account for roughly how much of the death spells are actually life stealing spells.

Now that isn't precise enough.

We want a wizard who has 8 chaos books and 4 death books to spend more on chaos spells during combat than death; and similarly, a wizard who has 7 death books, 1 life, 1 nature, 1 chaos (which is actually my current game), wants to spend far more on death than the other 3 combined.

So we change the formula to [casting skill]*[number of death books]/[total number of spellbooks] / [some number to roughly approximate how many death spells are also life stealing spells].

So for my 8 chaos 4 death wizard (always assuming 200 casting skill), they would have 67 base casting (200*4/12) skill for death spells, divided by how many of those spells will be life stealing.
The 7 death 1 life 1 nature 1 chaos wizard would have 140 base casting skill (200*7/10) for death spells, divided by how many of those spells will be life stealing.

(Actually incidentally this formula should be used for basic strategic combat formulations as well; simply having more books shouldn't inherently increase how much damage your spells are doing. The wizard with 200 casting skill and 8 chaos books should still only get 200 casting skill; another wizard with 14 chaos books but still only 200 casting skill shouldn't be getting 75% more 'casting skill' damage in strategic combat because of it.)

As for that divisor to approximate how many death spells are also life stealing spells (what you currently have as 5).. i'm inclined to agree that 5 is probably good. Strictly, I'd like to base it on tier (as you mentioned rare, although also very rare, doesn't add much to undead creation), which would mean approximating it based on current casting skill, but I believe that would be too complicated.

Not sure about that one. At 10 books and 100 skill the current result is 200 value - you get to convert an uncommon. 100 skill casts Syphon Life 3 times, or Life Drain 10 times, so that is realistic.
Reducing it to 100 (no uncommons, only 1 common or 2 if they are really weak) would be harsh. Even most decent normal units are over 100 cost.

Modifying the book thing as I describe above accounts for this. Dividing by 10 was to simplify it without bothering to actually account for it.

Of course, halving both, if we think it's too much, is an option.

Right, I think the ratio is right. And I think it should be halved overall. Except AI suck at getting undead out of tactical; in late game this doesn't matter (humans also have the problem of having a hard time getting undead in late game), but in early game it does matter. So I want to artificially increase their early game strategic undead by simply leaving it where it is right now to account for that. And it happens that create undead is an early game ability. I'm not suggesting making them different because it isn't working correctly, I'm suggesting it because the broken formula right now, actually works very well for balancing out the early game tactical problems a death AI has. (And in early game, the casting skill doesn't matter much, so the change I'm suggesting for casting skill won't impact early game anyway.)

[Bahgtru]

No, it wasn't missed, I was just really tired of this. At this point this discussion has probably taken up more of my time than the entire implementation of the building queues system which was supposed to be mythical tier difficult.

Well, not for me, I've done maybe 4 posts and I'm usually quite synthetic.

Ok - the HP point of reference actually makes sense. I'd still like to ask for the possibility to turn a werewolf though, I've been using them in the current version and it's really fun. With sleep not making werewolves, it shouldn't be a big issue given their high hp, and the decision to make them death immune IIRC stemmed from the sleep problem.

[Seravy]

Yeah I guess it feels so endless because I had to tackle multiple problems at once (wolves, ghouls, zombies, sleep, whatever) which were all related. Just this zombies/sleep thread by itself is already 6 pages long.
At the moment I'm happy I was able to reach a decision for all of them, even though it took, what, 3 weeks? 4?, to finally do it. At the moment I believe Illusion Immunity removal might even work better, and we haven't even tested a single game with them yet, I'm not going to overturn that decision for a vague, "yes, they are probably not that OP anymore". Ghouls are still a thing btw and would convert wolves with ease, even if Life Drain wouldn't be very effective against that much health. I prefer such cans of worms to be kept closed.
More importantly Lycanthopy and Blood Lust are easy researchable early Death spells - you can have your fun werewolves using your own spells, no need to get them from the enemies. In fact, using undead werewolves without having to research the two spells, stealing them from the enemy and making them do the work for you, is anything but fair or balanced.

...also it's bad design when a problem the player has to solve, ends up solving itself, and death wizards summoning the wolves they are weak against once turned undead, is exactly that sort of thing. Like how you don't make monsters weak to poison drop poison daggers unless your goal is to make an easy, braindead game.

[Nelphine]

I'd agree with that. The amount of life drains it would take to make a werewolf.. then again, ghouls.

How hard is it to get a werewolf with ghouls? If it's too easy, that still ends ups with the ai making werewolves, that the human then collects, and then the AI who made the werewolves has defeated itself.