Caster of Magic II Bug Reports!

[Seravy]

2020-10-06
-Added Roots of Genesis global enchantment image (by Slingers)
-Unit overdamage after losing a hp buff will now disappear, leaving the unit at 1 hp. (temporal solution. After the current test game I will store this damage in a new variable and make that restore the damage once it is healed overland or enters combat, whichever happens first.)
-Fixed bug : Negative food causes population growth instead of shrinking
-Fixed bug : Death Immunity doesn't protect from Cloak of Fear
-Astrologer retort will now show up in the combat info window during node battles.
-Settlers are now excluded from the diplomacy messages that refer the player's most powerful unit.
-Destroyed buildings now show "destroyed <name>" when mousing over them.
-Fixed bug : Defeat animation and undead raising window plays at the same time.
-Fixed bug : Defeat screen cannot be clicked away

Mostly minor stuff today but that's always good news. I'm playing a 4 wizard test game right now and so far AI performance feels roughly the same as CoM I. It's still the first half of the game though.

[massone]

Ended my first game (Lunatic, Maximal, Large Continents, 1.5 Power, Normal Climate, Rich Minerals) early at turn 75 with these observations:
-No serious bugs found yet aside from the negative health issue reported previously, which is great.
-When Engineers build roads, the right side infobox always seems to say "Unreachable!" instead of displaying the number of turns to build.
-Halflings can't seem to build the Monument line. Was that a deliberate design choice?
-Trade Brokering couldn't happen yet because I only met 4 wizards. The distances involved are massive, and it will take quite a lot of time to explore beyond them. Within this group, I had some trading options, but they were insignificant--I didn't actually trade anything.
-Many AIs are casting Tree of Knowledge in 1504.
-It appeared to me that some AIs had terrible starting positions, which may have caused problems for them when they couldn't easily cross the vast ocean distances. The new Spymaster doesn't show their split ratings anymore, so I couldn't be sure, but it seemed to me that their "Overall" power graph wasn't going up as fast as I remembered from COM I.

Restarted with "Wet" to have a better basis of comparison, as I usually played Wet before. New observations, currently at turn 60:
-AIs again casting Tree of Knowledge before 1505 compared to the Turn 72 restriction in COM I.
-Power graph varied greatly between AIs, not yet conclusive on their performance
-Still large distances between continents
-Several neutrals were near me, which made my start much stronger than the last game, despite having a 10-pop capital this time. There was a discussion about the impact of neutrals on gameplay, and it definitely does change the growth curve dramatically. Still, the luck of the map generation is part of the game, and I don't think it's unbalanced.

I intend to continue this second game much longer to see how spell trade develops. So far it looks to be irrelevant again at the early Uncommon stage, as I'm researching the most important spells myself. Additionally, AI Wizards seem to start on a lower relations threshold, and I'm not inclined to trade away important economy spells so early in the relationship. Finally, it would appear that the initial settlement rush has been extended significantly due to Maximal distances and extra land available, to the point where it's extremely easy to get at least 10 towns, and I expect there's not even a need for war to expand to 20+ towns.

Is the overextension formula still the same as before, with Maximal being +75%? It looks like I could massively outpace the formula just by settling, which means keeping positive relations and allies by staying under the limit could be a serious tradeoff of far less territory.

[Seravy]

-Halflings can't seem to build the Monument line. Was that a deliberate design choice?

Yes, it's a combat building and Halflings are supposed to be a peaceful race.
This might need to change when we update the race but that's something to worry about next year.

-Many AIs are casting Tree of Knowledge in 1504.

It appears on turn 54 the earliest which is 1504 July. It has a low research and casting cost so assuming the AI did get lucky enough to finish a research at that time and don't have anything higher priority on the list it's possible considering the difficulty level, but still seems unlikely. You can see the AI's research choices and casting choices in the log file as well as spell treasure and turn count so if there is a problem (spell being available too early, or the AI casting it without having the spell) that should be visible from the logged data.

The date is set earlier because there are three layers of additional delay - the player has to finish their previous research to make a spell appear, they need to research this spell, and then they need to cast it. The latter two might not take much time for a Lunatic AI but otherwise it's 3-4 turns each.

About the power graphs, I'm a bit worried it might be overrating magic power. (You can see the separate bars if you click the button that shows "overall" to cycle through the charts)
Hard to tell because army strength vastly changes once people start building better tier units while magic power scales a bit more linearly. Same for population as expanding number of cities make it exponential. So I think that inherently makes magic power overrated early on but it's also the most important factor at that portion of the game so it might be ok as is.

Otherwise, chart fluctuations might be the result of starting positions and available nearby land area. Even in CoM I, AIs that started on islands, near tundra, too much deserts, small continents, etc performed much worse than AI that started on larger land, near rivers, etc, and while the fortress itself has has to be an at least average land, the other cities beyond that might not be good. That's not necessarily a bad thing as it allows for player choice - fight the easy, weak enemy or the strong one that will snowball into unstoppable if left alone. It also helps AIs to beat each other.

The overexpansion formula is

Code:

toomuchcities:= (3+GD.Landsize)*(6+Gd.CurrentTurn div 12) div 6;

So I guess Normal land is 3+3 = 6, Maximal would be 9, a +50%.

This doesn't exactly match the difference in land available which is +100% but considering producing the additional settlers take additional turns, it looks fair - time allows more cities.

On the old Normal size maps there was about 10-12 city spots per player so 20-24 on Maximal sounds reasonable. Of course if you outpace the AI at building settlers then there is 10 other player's worth of territory to hog for yourself instead of 3 and starting locations might affect how much you can use easily.

Yes, the warning will trigger if you build settlers too fast, that's what it is for. If you build all 20 cities before 1407 they have all the reason to be angry, not even the AI spreads that fast. (at least not on the average difficulty levels. Lunatic might be different but I doubt. Settlers take a while to reach their destination even if the AI builds them faster.)
...but chances are rampaging monsters will prevent that from happening because more territory spawns them more often.

Nice to hear spell trading wasn't relevant. I was hoping for that. (but we do have the trade value penalty already included so maybe that's part of it. At max players your spells are worth 4 points less in trade than on 4 players.)
Also, land size definitely affects spell trading a lot. I was able to discover all wizards on my plane by mid uncommon. (though I was playing Expert so mid uncommon was later than on Lunatic.)

[rgirardo]

Hi Seravy[/url][url=https://www.realmsbeyond.net/forums/member.php?action=profile&uid=6122]
I wanted to know if you still needed testers for the Caster of Magic 2, I am a huge fan of your work and love the original Master of Magic I have been playing it for many years.  MOM is the best game along with MOO2.  I like to try and win the games with no saves on impossible. 

I know that you have heaps of testers, but would you like 1 more to help out?

[massone]

It appears on turn 54 the earliest which is 1504 July. It has a low research and casting cost so assuming the AI did get lucky enough to finish a research at that time and don't have anything higher priority on the list it's possible considering the difficulty level, but still seems unlikely.

Just checked the log, and I had:
-5 AIs start researching it on turn 54, finish and cast instantly on turn 55. Apparently I also started 54 and finished research on turn 55, but unlike the AIs I didn't bother casting it because I didn't need it--I was just clearing out easy to research spells from the book to make the Uncommon I wanted appear.
-almost everyone else seems to have selected it and finished cast by turn 61.

This is Lunatic so AI being faster at it is to be expected, but I wonder if the early rush races like Barbarian who don't have advanced units might suffer for it. Especially on a Maximal map like now, by the time Tree of Knowledge has been cast, there's hardly been any time to engage an enemy in war. The only wizard I can reasonably go to war with right now has one town on a continent we're both settling which he conquered from a neutral, and another small city he settled 15 squares away on another large continent (large enough to fit like 10+ towns) which is otherwise empty, and the rest of his towns are on yet another continent many squares away that I haven't even discovered. The other 3 wizards I've found are all at least 20 squares away from my nearest border town in 3 different directions. We can't really attack each other even if we wanted to, it's just too far.

About the power graphs, I'm a bit worried it might be overrating magic power. (You can see the separate bars if you click the button that shows "overall" to cycle through the charts)

Thanks, very useful. It doesn't seem like magic power is being overrated in my current game. The overall, military, and magic power are all pretty much aligned for the 4 wizards I've met so far, and me.

The overexpansion formula is

Code:

toomuchcities:= (3+GD.Landsize)*(6+Gd.CurrentTurn div 12) div 6;

So I guess Normal land is 3+3 = 6, Maximal would be 9, a +50%.

It looks like the threshold is slightly lower than the old formula, which makes sense if there's more wizards. Right now I have 14 towns turn 68, and apparently two of the Lunatic AIs I've met have expanded even more than me based on the population graph. I was actually planning to settle 3 more towns before the end of 1505, until the AI declared war on me, so it seems likely that at least the AIs will exceed 20 in 1407.

[Seravy]

Today's update :

2020-10-07
-Added event image for Atlantis (Slingers)
-Fixed bug : Heroic Heart had no sound effect
-Road building will no longer show "unreachable" when the road destination is the same as the start.
-Road building will now show the expected number of turns to complete the road (assuming no more than 1 tile moved per turn)
-Fixed bug : Units not selected in the stack count towards enabling the Build/Purify/Meld/Settle button.
-Fixed bug : stat penalty icons are still wrong.
-Added logging of unit attack targeting simulated damage and suppresson modifier.
-Increased accuracy of attack damage simulation (100 iterations instead of 20)
-Increased suppression effect on AI combat attack targeting priority. (15->25% modifier)
-Fixed bug : combat unit stat display shows magic weapons on units that don't have actual (+1 to hit) magic weapons, only the ability to bypass weapon immunity due to a buff or being fantastic.
-Improved unit and item enchantment aura formula
-Fixed spell charges not displaying the dot in front of the text
-Added logging of Hostility changes due to hostility reevaluation
-Fixed bug : Combat move matrix is not refreshed for the selected unit after casting a spell.
-Fixed bug : Corruption and Raise Volcano was cast on targets without hostility by the AI
-Fixed bug : the AI can't use curses at Annoyed hostility level.

Test game recording until today : https://youtu.be/yD19W-95AUM

[massone]

Further observation has shown that my previous thought that AIs not growing as fast as I remembered was wrong. Some of the AIs are wildly outpacing the others because of node captures. On turn 77, it seems two of the AI's I've met appear to have 650+ Power income based on the graph, as their line on the graph is 4 times higher than my own (173), and continuing to go up almost vertically, whereas two others I've met only have about ~200-250. Our population curves aren't very far apart, so the difference must come from nodes.

One AI in particular has two doomstacks of 9 gargoyles moving to take out nodes on a continent which only I have settled, and apparently doesn't even bother garrisoning it, leaving it for me to take, I can only assume because the AI already has tons of other nodes in areas I haven't explored yet and they're just attacking faraway lairs for the loot because they have a giant army of summons already.

Also, these gargoyle doomstacks seem to be taking out lairs/nodes with 0 health losses. That should not be possible against a node with many Phantom Beasts that can ignore armor or even a few Fire Giants. Is this supposed to happen with the new quick combat calculations? I was under the impression that you'd changed it to concentrate damage on each unit instead of spreading it out. Even so, they should be taking some damage each battle either way.

These AIs are now taking out nodes much earlier than they used to in COM I, and it's discouraging that they can seemingly make a beeline to all the lairs they know they can defeat without exploring anything, and apparently without any health losses whatsoever despite not having any healing spells. They don't have astrologer either, one of them even has no retorts at all.

EDIT: Forgot to mention that I'm still using the game build from a few days ago when I first started. Wasn't sure if the updates were save compatible so I didn't update yet.

[Seravy]

Further observation has shown that my previous thought that AIs not growing as fast as I remembered was wrong. Some of the AIs are wildly outpacing the others because of node captures. On turn 77, it seems two of the AI's I've met appear to have 650+ Power income based on the graph, as their line on the graph is 4 times higher than my own (173), and continuing to go up almost vertically, whereas two others I've met only have about ~200-250. Our population curves aren't very far apart, so the difference must come from nodes.

The charts do include the bonus for difficulty levels, which on Lunatic should be something like 2x-2.25x power for the AI. So that's half the difference but the other half must be the nodes (and more developed cities).

One AI in particular has two doomstacks of 9 gargoyles moving to take out nodes on a continent which only I have settled, and apparently doesn't even bother garrisoning it, leaving it for me to take, I can only assume because the AI already has tons of other nodes in areas I haven't explored yet and they're just attacking faraway lairs for the loot because they have a giant army of summons already.

It's because the doomstack took it and those are not allowed to stay behind to garrison nodes. If the AI doesn't have nearby troops to pull onto the empty node, which is likely because it needs land units - water units would eventually go into forming a new doomstack - it stays empty.

Also, these gargoyle doomstacks seem to be taking out lairs/nodes with 0 health losses. That should not be possible against a node with many Phantom Beasts that can ignore armor or even a few Fire Giants. Is this supposed to happen with the new quick combat calculations? I was under the impression that you'd changed it to concentrate damage on each unit instead of spreading it out. Even so, they should be taking some damage each battle either way.

The damage is concentrated only for the neutral (lair) side. If you have a save where it happens, look at the log, all combat damage and result is logged. Flight might be causing this - nonflying units cannot hurt flying units because retaliation damage isn't a thing in automatic combat. So any damage they manage to do before their attack power is dropped to zero to simulate they are avoiding melee to stay alive will weaken the lair as well as any damage from AI spellcasting power. The AI learned to go for combats that end up with their army retreating exhausted if the losses on the other side are much higher, so they can in theory clear lairs by attacking it repeatedly using flying enemies like human players do with sprites or gargoyles.

The updates the past few days were all save compatible. I won't change save format until my own test game is over.

[massone]

If you have a save where it happens, look at the log, all combat damage and result is logged.

When does the log file update? I reloaded 3 times and to replay the turn when it happened and end turn, but the log file continued to show the wrong turn's info from before the reload. I had to close Caster and delete the log file to get it to refresh.

Anyways, this seems to be the battle log:

Code:

Stack of 9 units moves from 1,76,108 to 73,111
Stack arrives to 73,111 plane :1
Stack initiates combat!
Entered combat at X=73, Y=111P =1
Combat turn 1 player 1
Combat turn 1 player 2
Casting uses 0 MP
Buff = 0
Buff Ranged = 0
Damage = 0
Curse = 0
Summon = 0
Combat turn 2 player 1
Combat turn 2 player 2
Casting uses 0 MP
Buff = 0
Buff Ranged = 0
Damage = 0
Curse = 0
Summon = 0
Combat turn 3 player 1
Combat turn 3 player 2
Casting uses 0 MP
Buff = 0
Buff Ranged = 0
Damage = 0
Curse = 0
Summon = 0
Combat turn 4 player 1
Combat turn 4 player 2
Casting uses 0 MP
Buff = 0
Buff Ranged = 0
Damage = 0
Curse = 0
Summon = 0
Combat turn 5 player 1
Combat turn 5 player 2
Army deals 45 Antifly Melee damage
Combat turn 6 player 1
Army deals 52.4603107887445 Melee damage
Combat turn 6 player 2
Army deals 28.0655703950022 Antifly Melee damage
Combat turn 7 player 1
Army deals 42.3467360082209 Melee damage
Combat turn 7 player 2
Army deals 19.5400477012301 Antifly Melee damage
Combat turn 8 player 1
Army deals 35.3053778294973 Melee damage
Combat turn 8 player 2
Army deals 14.5913185703337 Antifly Melee damage
Combat turn 9 player 1
Army deals 30.0473200230616 Melee damage
Combat turn 9 player 2
Army deals 11.4462690924969 Antifly Melee damage
Combat turn 10 player 1
Army deals 25.9225972115994 Melee damage
Combat turn 10 player 2
Army deals 9.31778865230631 Antifly Melee damage
Combat turn 11 player 1
Army deals 22.5648834363969 Melee damage
Combat turn 11 player 2
Army deals 7.80953929103979 Antifly Melee damage
Combat turn 12 player 1
Army deals 19.750675190858 Melee damage
Combat turn 12 player 2
Army deals 6.70308223855759 Antifly Melee damage
Combat turn 13 player 1
Army deals 17.3351845193509 Melee damage
Combat turn 13 player 2
Army deals 5.86953500094336 Antifly Melee damage
Combat turn 14 player 1
Army deals 15.220066945631 Melee damage
Combat turn 14 player 2
Army deals 5.22869817665086 Antifly Melee damage
Combat turn 15 player 1
Army deals 13.3358782664028 Melee damage
Combat turn 15 player 2
Army deals 4.72849966422946 Antifly Melee damage
Combat turn 16 player 1
Army deals 11.6319387319196 Melee damage
Combat turn 16 player 2
Army deals 4.33394910008666 Antifly Melee damage
Combat turn 17 player 1
Army deals 10.0701775525881 Melee damage
Combat turn 17 player 2
Army deals 4.02087428372271 Antifly Melee damage
Combat turn 18 player 1
Army deals 8.62123451743229 Melee damage
Combat turn 18 player 2
Army deals 3.77220785686321 Antifly Melee damage
Combat turn 19 player 1
Army deals 7.2618997271136 Melee damage
Combat turn 19 player 2
Army deals 3.57570310210903 Antifly Melee damage
Combat turn 20 player 1
Army deals 5.97337645083022 Melee damage
Combat turn 20 player 2
Army deals 3.42248554386681 Antifly Melee damage
Combat turn 21 player 1
Army deals 4.74006592138087 Melee damage
Combat turn 21 player 2
Army deals 3.30611228289646 Antifly Melee damage
Combat turn 22 player 1
Army deals 3.54869110393425 Melee damage
Combat turn 22 player 2
Army deals 3.22195087401694 Antifly Melee damage
Combat turn 23 player 1
Army deals 2.38764428945916 Melee damage
Combat turn 23 player 2
Army deals 3.16676654826016 Antifly Melee damage
Combat turn 24 player 1
Army deals 1.24648343480597 Melee damage
Combat turn 24 player 2
Army deals 3.13845068801691 Antifly Melee damage
Sss'ra spent 0 mana!
Defenderlossrate Neutral=1
Attackerlossrate Sss'ra=0
Defeated : Neutral's Phantom Beast died in battle!
Defeated : Neutral's Phantom Beast died in battle!
Defeated : Neutral's Phantom Beast died in battle!
Defeated : Neutral's Phantom Beast died in battle!
Defeated : Neutral's Phantom Beast died in battle!
Defeated : Neutral's Phantom Beast died in battle!
Defeated : Neutral's Phantom Warrior died in battle!
Defeated : Neutral's Phantom Warrior died in battle!

It does appear as if the Phantom Beasts did nothing to the Gargoyles because they were flying. It seems like a pretty lopsided outcome for what would otherwise have been a very difficult battle. Even a human player abusing the retreat tactics would've suffered huge losses, and/or spent a lot of mana over many turns. Here, they swept through and moved on spending 0 mana, didn't have to heal or wait for anything. I don't have the time to dig through the log for the other nodes I saw them defeating, but I think there's something odd going on here even without the Flying issue. It could also be a matter of there being more nodes in general, especially outside of controlled territories, so the AI is able to attack more, but the pace at which the AI has been able to capture nodes is much faster than it was in COM I, even before the nerf to Lunatic bonuses. In my old games, the AIs typically captured just 1 or 2 nodes by turn 72, but here it's looking like they've got 2-3x more melded, and still more just defeated but not melded.

[jhsidi]

Haven't had a chance to test in a few days, but in at least two of my previous test games, AI gargoyle stacks specifically seemed to run rampant.

As a player, gargoyles aren't that powerful -- in particular they're extremely weak against phantom warriors / beasts. But for AI they're pretty overpowered with the auto combat setup.