Frogboy's custom FE 1.29 build » Forum Post by Frogboy

Frogboy

Join Date 03/2001

+2419

@Mytheos, what OS you running?

Successfully updated karma reason!

Frogboy

Join Date 03/2001

+2419

When FE crashes, it does send us a crash report. The 1.29toad build has crashed a total of 4 times. Ironically, each time is on a different version of Windows (which implies 4 different users).

Successfully updated karma reason!

Alstein

Join Date 07/2004

+55

Quoting Frogboy,

reply 52

When FE crashes, it does send us a crash report. The 1.29toad build has crashed a total of 4 times. Ironically, each time is on a different version of Windows (which implies 4 different users).

It's possible that 2 of those crashes were me on 2 different laptops .

To be honest, my reasoning for saying it's improved is that the game doesn't feel like a cakewalk the way 1.2 did on fair settings.

Successfully updated karma reason!

Frogboy

Join Date 03/2001

+2419

The population thign with pioneers is really nerfing the AI.

Successfully updated karma reason!

Alstein

Join Date 07/2004

+55

Quoting Frogboy,

reply 55

The population thign with pioneers is really nerfing the AI.

I suspected that was the culprit given how much easier the game was despite sensing general AI improvements.

Is it a lack of strategies to work around this limitation that is holding you back, or is it just harder for the AI period.

I do think the AI's with arcane monolith need to be using it more. Also, maybe AI's with good production in their first city should use Sovereign's Call in early game, than cancel once they get the city growth up.

Successfully updated karma reason!

Frogboy

Join Date 03/2001

+2419

No, the issue was I didn't know that was going to be changed until after released. So there was no AI code to take that into account.

Successfully updated karma reason!

Hey Brad, I'm one of the guys who was crashing a lot before but I haven't crashed (or had save game corruption) with 1.29frog build on my Windows 8 machine... haven't tried my Windows XP machine yet.

I think the AI is doing great on Expert and Ridiculous, but I'm steamrolling it on Hard. I'm playing standard Gilden (no customization). I think the AI is rushing pioneers when it has enough gildar (on Expert and Ridiculous), but can't generate that kind of income on Hard.

By the way, one thing I've noticed is that if I summon an Earth or Fire Elemental my power rating goes way up... I mean way up... and I take over the lead and then everybody cowers before me. I don't think that should happen.

One last thing, when Resoln surrendered to me, the sovereign still had too many greaves, was not riding her warg, and was carrying two swords that were better than the one she had equipped. FIX IT!

Successfully updated karma reason!

Frogboy

Join Date 03/2001

+2419

I'm working on a new build here. If you have any quickie specific AI stuff I can put it in.

I just got a new diplomacy (internal to the AI) thing in that I think will make the AI behave..better.

I'm running it through some stability testing scripts now to make sure it's all really solid.

IF a unit isin't riding a mount of some kind by mid game, that's a bug. I've made the AI go and at its earliest ability buy mounts if it doesn't win one on a quest.

+1

Successfully updated karma reason!

Alstein

Join Date 07/2004

+55

One common sense switch that's diplomacy related- why is it that AI's will surrender more quickly then they sign peace treaties? That should never be the case.

I don't mind the AI being stubborn in accepting peace- usually when a player asks for peace it's because they don't want to fight more.

Suggestion #2- on running AI, an AI unit in tactical shouldn't run away, if it's going to be shot to death in the process, it should kamikaze attack instead.

Suggestion #3: If an AI has a shot to avoid attacking a cloak of fear unit to attack another unit (imp of the titan specifically), it should.

Suggestion #4: spiders shouldn't web/beguile ranged units. This has been asked for at least 2 months now.

Successfully updated karma reason!

Quoting Alstein,

reply 60

Suggestion #4: spiders shouldn't web/beguile ranged units. This has been asked for at least 2 months now.

I think a better answer is that webbing should cause range units to lose their attack. If my spiders use web on AI archers, they lose their attack. The AI should be able to use web the same way.

Successfully updated karma reason!

Alstein

Join Date 07/2004

+55

Losing attack is too harsh, maybe have it -50 accuracy.

Successfully updated karma reason!

Don't forget about imprison, it works the same way as webbing.

Successfully updated karma reason!

Frogboy said in another thread that he's fixing the webbing of archers. He didn't say he was doing beguiling because he was responding to my post and I didn't think to include beguiling, but he might be doing that as well.

Successfully updated karma reason!

BlackSmokeDMax

Join Date 03/2006

+60

I like the idea of ranged units losing attack, but maybe just for one or two rounds if webbed. Think of it as time spent cleaning up the web goo.

Successfully updated karma reason!

Found a bug, I think it's a general one (pretty sure I made it happen in another patch too) but I'm running your build so I'll post it here.

A 20 stack of AI troops invade my territory. I firestorm them and my sovereign dies? Odd behaviour to say the least. It's reproducable in the save for me at least, even with Sovereign stuck inside a city when casting it.

https://dl.dropbox.com/u/95181126/Firestorm%20Bug.EleSav

Oh, and dunno if that has something to do with it, but that wildlands fire power thing (+3 fire power) was bugged in this game too, and I could build it twice.

Successfully updated karma reason!

Thanx for the update Brad. Your 1.29 patch has allowed my aging XP machine to get past turn 100, and I have regained use of all of my previous saves. Good job. Your efforts are appreciated.

Successfully updated karma reason!

MarvinKosh

Join Date 08/2005

+87

Quoting Darxim,

reply 64

Frogboy said in another thread that he's fixing the webbing of archers. He didn't say he was doing beguiling because he was responding to my post and I didn't think to include beguiling, but he might be doing that as well.

When the tactical map is fairly long, stagger the web attacks. Wait for player units to try and cross to where the spiders are. That way the AI has a chance to swarm around the one or two enemy units that managed to resist. If there are any web attacks spare after doing this, save them for re-webbing player units that break their webs.

When you consider that in the expansion there's going to be a bonus specifically for using this tactic, you'd have to be crazy not to.

Successfully updated karma reason!

I would say too that the AI is better and smarter, but weaker perhaps because of pioneer thingy.

Two things I noticed that felt wrong:

Strategic AI: The AI sometime move around an enemy outpost, probably because its goal is further away, whereas the more sensible move would be to move over the outpost and then proceed. Perhaps a test on each stack, if adjacent to an enemy outpost (so no pathfinding involved here, a real time save), switch the first movement leg to take the outpost then continue to previous goal.

Tactical AI: AI hero, alone, facing 6 units of pikemen and my fire archmage sovereign cast ... Protection from cold (and I had no possibility to send cold spells either!).

More specifically, 'protection from' should never be cast randomly but should make sure the enemy has some significant damage dealing possibility in said element.

easier said than done, I know.

Successfully updated karma reason!

They should web poisoned units if possible. BUT move away from harm first.

Successfully updated karma reason!

I really appreciate the AI work...

But there seems to be 3 fundamental things the AI (or its coder ? ) does not seem to understand correctly about the early game (expert/expert) :

- AI neglects early city spam: since the game has no mechanisms to check expansion (corruption, happiness, whatever), early city spam is essential. I see too many idle settlers and too many outposts being created while city real estate is available, even prime, and even nearby. Outposts are close to worthless compared to cities (no self defense, no production, no tech): no outpost should be created until even the most worthless piece of real estate has been populated. And no settler should be left idle.

- AI neglects heroe recruits: also a top priority, AI lets me recruit most if not all of them and kill those of the opposite team.

- AI neglects ancient libraries: these are the top priority goody huts in the early game, the AI should be homing in on them like if there was no tomorrow. I still see unexplored libraries deep inside AI territory, far from my own.

Keep up the good work

Successfully updated karma reason!

Alstein

Join Date 07/2004

+55

the ancient libraries seem to be only +1 research. Nice, but not as important as clay pits, and no more important than most other improvements.

Successfully updated karma reason!

Quoting ulysses_31,

reply 71

- AI neglects early city spam: since the game has no mechanisms to check expansion (corruption, happiness, whatever), early city spam is essential. I see too many idle settlers and too many outposts being created while city real estate is available, even prime, and even nearby. Outposts are close to worthless compared to cities (no self defense, no production, no tech): no outpost should be created until even the most worthless piece of real estate has been populated. And no settler should be left idle.

I think that is an issue, but the bigger problem is the balance of horizontal vs vertical expansion combined with the cost vs reward of outposts. I would do something to buff outposts like decrease the build time on improvements with possibly some buffs also and give them higher base area of control. I think the changes to food, growth, and unrest in the expansion will alleviate the settler costing population burden, but that doesn't help us now does it

and i think the ancient libraries he was talking about are the ones that give you the one time +20 research when you get to them on the map. they are really a nice boost.

Successfully updated karma reason!

Alstein

Join Date 07/2004

+55

Agreed on both points above. Vertical expansion needs to be rewarded more in general- that's a general problem with this genre- it always seems to be underdone.

That said, I suspect weak vertical expansion helps the AI- is that true as a general rule?

Successfully updated karma reason!

Quoting Alstein,

reply 72

the ancient libraries seem to be only +1 research. Nice, but not as important as clay pits, and no more important than most other improvements.

I am talking about the +20 or +100 single shot ones, not the resource types (which I prefer to clay pits, BTW: early tech trumps production)

Successfully updated karma reason!

Quoting Alstein,

reply 74

That said, I suspect weak vertical expansion helps the AI- is that true as a general rule?

I would think humans are better at both (vertical + spamming), but the AI can (and does) cheat at higher levels to counter this.

The issue with FE is that there is no balancing mechanism to counter city spam benfits + the cost of buildings is so high (both techwise and production).

Successfully updated karma reason!