Anyone Else Go Back to FFH2 version 22?

[Sarisin]

The last few weeks I have been enjoying playing version .22H games.

Yesterday, I decided to load up .23 again and see what happens. After losing in short order 5 battles where I was heavily favored (Combat Odds 90% or higher) I am back to .22H again.

I know the developers are busy with Shadow and other things, but has any thought been given to solving the combat problem described at length in other threads?

Would it be that difficult to go back to the unit strengths found in version .22? I'm far from being an expert, but it seems like when the strengths increased in .23, but combat (attacking and defending) problems started.

If it is too much, that's fine as I am happy playing the mod using the older version, but a little less apt to participate in this forum as I feel most are using the new .23 version.

Anyone else go back to .22?

 

[TheJopa]

I never noticed any difference with odds in 0.23 and 0.22. Losing with high odds was in since... first release of FfH and first version of civ4.
Anyway I would never, ever go back one version, features you lose take away so much from the game

 

[Calavente]

I will wait to end my current game before going to 0.23 ... I didn't made the switch due to alarming information considering combat odds..

if I find it borked then, I think I will try the smarter orc mod, iirc the modder found a solution for the combats, or maybe not...

 

[vorshlumpf]

I still don't see a problem with combat in .23, other than the discrepancies in combat odds display.

 

[gandalf51]

Sarisin try Smarter Orcs mod

 

[Pocus]

Played 460 turns of a huge game in 0.23c. Nothing special to report about combats.

 

[Vulcans]

i haven't noticed anything strange with the combat odds in 0.23

naturally you can still occasionally lose when you have 90% odds, just like in Civ4. that's the thing with probability, it brings in an aspect of luck.

i do know that the exact odds of winning are never the same as the displayed ones, same applies with civ4. as the damage done per hit changes it will mean that you'll need a different integer of battle wins to deal a lethal blow as described in combat mechanics in the war acadamy.

 

[daladinn]

try running vanilla civ 4 and tell it to patch. after patching vanilla all my combat issues went away. it could be issues between 23 , vanilla , and dir x.

 

[Kael]

Their were a few significant combat odds problems that were discussed heavily. Resists were being calculated incorrectly, and defensive percent promotions weren't being maintained through saves. Both of those issues existsed in 0.22 and 0.23 up until 0.23c.

edit: incidently player that attack more than they defend may notive that 0.23c seems significantly harder than prior versions. The reason is that since the defensive percent bonus is being lost (in all "Fire" version prior to 0.23c) when the game is saved and reloaded it puts defenders at a large disadvantage. So although combat may seem much harder in 0.23c for attackers, its not that there is a combat problem in that version, only that combat is actually working correctly in that version.

 

[loki1232]

If i use that "increase all hp" trick, does it break saves?

 

[Sarisin]

Thanks for the replies.

I compared the losses when heavily favored >90% odds with .23C, .22H and BtS. I use pretty much the same game variables for all three (map size, game speed, etc., raging barbs, etc.).

I'm sorry, but it just isn't even close.

Yes, I realize it is possible to lose a battle you are heavily favored to win. History tells us this too.

However, I'm certainly there are far more of these losses in .23C than in the other two.

Initially, I just started quick saving before nearly every battle in .23C, but that both slows down the game and takes away the 'fun' surprise element. There is no fun surprise element when you lose five >90% combat odds battles in short order.

If you go back and read, I believe, the 'Balance' thread you will find that this was, and IMO, still is a significant problem in .23C.

Maybe the game is just catching up in terms of difficulty. Fine, I have no problem with that. But, then, the Combat Odds really need work. If the odds REALLY are 89% vs. the 99% that is showing, then I would reconsider making that attack with a unit I don't want to lose.

Or, maybe it is just best to ignore the Combat Odds completely if they don't reflect anything accurate.

Again, though, I just haven't seen this in .22H or BtS. I have played quite a few of the former games and several BtS games, and the odds seem pretty accurate. You just don't lose as many battles you are heavily favored to win.

I want to stress that I am not complaining here at all. I enjoy using .22H and I don't believe there are that many real differences between it and .23.

Thanks for listening.

 

[OzzyKP]

I suppose this issue has been discussed to death and I suppose (hope) it'll be addressed with BTS (which has a better odds calculator built in regardless) but I'll just add my voice to the chorus saying how frustrating it is to consistently lose battles with 95% odds.

 

[xanaqui42]

Their were a few significant combat odds problems that were discussed heavily. Resists were being calculated incorrectly, and defensive percent promotions weren't being maintained through saves. Both of those issues existed in 0.22 and 0.23 up until 0.23c.

Resistances are being calculated incorrectly in 0.23c (and I think all versions since resistances were put in). I have a detailed post, largely code impacting 20 functions, on this very topic. The easiest way to replicate the defect:

Enter world builder.
Give one player 3 warriors, each with Immune to Fire. Give another player 3 Fireballs. Attack the Warriors with the Fireballs.

Ideally, the odds should show 0% chance of success, and on the actual attack, none of the Warriors should be wounded (or at least their wounds should be very minimal - less than 6 HP on average, if the Fireballs are restricted to a minimum of 0.01 strength). If you try this with 0.23c, you will notice that this won't happen. If you try it with Smarter Orcs 0.11 or later, it will happen.

Note that the apparent "fix" in 0.23 to the combat odds in CvUnit::updateCombat is incorrect (the fix mentioned above fixes this too). This is an attempt to pass in a pointer to a defender where a pointer to an attacker is expected (and only if the unit is the defender); see the following line for an interface example:

int CvUnit::currCombatStr(const CvPlot* pPlot, const CvUnit* pAttacker, CombatDetails* pCombatDetails) const

So while this change in 0.23 does account for resistance in the combat odds, it neither accounts for them in combat, nor does it for the AI's combat odds estimation function. Even worse, because it's trying to treat a defender as an attacker, certain bonuses and penalties are counted twice, which causes a large deviation between combat odds and combat results in some circumstances (completely ignoring the resistance issue).

So although combat may seem much harder in 0.23c for attackers, its not that there is a combat problem in that version, only that combat is actually working correctly in that version.

I think that most complaints are about is a large difference between displayed combat odds and perceived combat results.

 

[Blakmane]

I think it's also got a lot to do with the nature of FFH as well. We all have stories of losing helicopters to swordsmen once in a while, but in vanilla it's not really an issue because units are just faceless pawns. However, in FFH when heroes and highly experienced units, which often fight a large number of battles, get unlucky then it's really noticable.

I think this largely an inherent issue with the Civ IV combat system that needs to be fixed.

 

[Arpymaster]

Last night I lost 3 different battles (even reloading the saves got the same results) with victory odds of 98% aprox (1 hunter lv5 vs a weakend giant, monks vs warriors in cities, etc).

I'm not complaining, though

 

[Bringa]

I think it's also got a lot to do with the nature of FFH as well. We all have stories of losing helicopters to swordsmen once in a while, but in vanilla it's not really an issue because units are just faceless pawns. However, in FFH when heroes and highly experienced units, which often fight a large number of battles, get unlucky then it's really noticable.

I think this largely an inherent issue with the Civ IV combat system that needs to be fixed.

Blakmane: I agree with you 100%. I just mentioned this to a friend earlier today when I described one of the aspects of Smarter Orcs.

In short, Vanilla civ is about historical combat with tons of faceless units. The combat may be random, but one unit won't be much different from the next, so if you lose unit a in a 90% combat while unit b wins a 10% combat, it all evens out. Not so in FFH: Heroes, national units, and stronger promotions all combine to make arbitrary combat all that much more frustrating. If you have an awesome hero but can't use him unless combat odds are 99% in your favor, that makes the hero feel really useless.

This is why I proposed a new combat system, which has already been implemented in Smarter Orcs. I'm sure our solution is still far from perfect, but I notice that I cannot play FFH without the new system anymore. It just feels wrong.

 

[calmon]

Last night I lost 3 different battles (even reloading the saves got the same results) with victory odds of 98% aprox (1 hunter lv5 vs a weakend giant, monks vs warriors in cities, etc).

I'm not complaining, though

As far as i understand simple reloading never change a combat result. You've to wait a turn or so to get another result

 

[Bringa]

Whether or not reloading changes the outcome of a battle depends on whether or not you selected "new random seed on reload" when creating the game. By default that option is off; if you leave it off, a combat will always end the same way, no matter how often you reaload. If you turn it on, you should get different results after each reload.

 

[Sarisin]

OK, as I have done in the past (e.g., spawning barb cities, and barb mounted unit withdrawal rates), I have run a small sample test.

Using Monarch, Huge Fantasy map, Epic, I played the first 250 turns of Arendel Phaedra/Elves. Raging barbs, so all my battles were with animals and barbs.

I had 68 battles where I was the attacker and the combat odds favored me by 90% or more. Of course, I defended many, many more times than this, but I always attacked when I had those >90% odds.

How many of these battles did I lose?

Eight. It included everything from a Scout losing to a Lion (96.7%) to Gilden Silveric losing to a common barb Lizardman (99.9% - doncha just love those?)

Honestly, I thought there would be more. But, still I believe it was more than you would expect in that number of battles.

I realize it was not a significant sample size, but I was getting sleepy.

Another thing I have noticed is the time between turns in FFH2 and BtS is quite different, with FFH2 taking much longer. Any idea why? Same game variables.

 

[Kael]

Resistances are being calculated incorrectly in 0.23c (and I think all versions since resistances were put in). I have a detailed post, largely code impacting 20 functions, on this very topic. The easiest way to replicate the defect:

Enter world builder.
Give one player 3 warriors, each with Immune to Fire. Give another player 3 Fireballs. Attack the Warriors with the Fireballs.

Ideally, the odds should show 0% chance of success, and on the actual attack, none of the Warriors should be wounded (or at least their wounds should be very minimal - less than 6 HP on average, if the Fireballs are restricted to a minimum of 0.01 strength). If you try this with 0.23c, you will notice that this won't happen. If you try it with Smarter Orcs 0.11 or later, it will happen.

Note that the apparent "fix" in 0.23 to the combat odds in CvUnit::updateCombat is incorrect (the fix mentioned above fixes this too). This is an attempt to pass in a pointer to a defender where a pointer to an attacker is expected (and only if the unit is the defender); see the following line for an interface example:

int CvUnit::currCombatStr(const CvPlot* pPlot, const CvUnit* pAttacker, CombatDetails* pCombatDetails) const

So while this change in 0.23 does account for resistance in the combat odds, it neither accounts for them in combat, nor does it for the AI's combat odds estimation function. Even worse, because it's trying to treat a defender as an attacker, certain bonuses and penalties are counted twice, which causes a large deviation between combat odds and combat results in some circumstances (completely ignoring the resistance issue).



I think that most complaints are about is a large difference between displayed combat odds and perceived combat results.

Great information, I'll check into it.