Combat Discussion

[Schwerpunkt]

I suppose proper Ev would be hard to balance. But I'm not sure I like the notion of glancing blows. Clearly they need to stack more hit.

A more practical solution is to instead look at evasion as mitigation -- dodge, block, parry. If an attack is dodged, it's a regular miss. If an attack is blocked, an amount of damage commensurate with the quality of the shield is reduced from the incoming damage. If an attack is parried, the defender gets +5 CT or something. Those would all be passive mitigation. For active mitigation, you'd have things like protect and shell.

Tanky types would trend towards block, physical types towards dodge, and probably everyone would get a small amount of parry (if hexes, frontal cone; if squares, full parry rate at the front and half at the sides). Parry would be weapon-driven, block shield-driven, and evasion would probably be derived from speed directly (something like Speed * 3, maybe).

Thoughts on this?

[Mordred]

Parry seems to be the kind of thing I'd expect from a Reaction ability rather than an inherent passive defense.

Block is exactly what I'm talking about with hit vs glance chance for armor.

Dodge is the problem the glance idea is attempting to address. I don't think it's a good idea to build classes around the idea of completely negating damage, because then you have to institute HoI's Evade caps, which puts an upper limit to how effective it is to buff evade. The whole hit-glance concept was meant to create an environment in which characters could stack EV to 100 without actually being untouchable.

[Ilium]

Why couldn't we drastically reduce current evade modifiers? Instead of being able to add an extra 10% M-Ev by equipping a piece of armor, its somewhere in the 2-5% range. Same with abilities. So with stacking we could reasonable only reach 20% additional evade in either category.

[Mordred]

If you just cut down on Evade as a whole without changing the way accuracy, armor, evade, and HP interrelate, all you do is make dodgers weaker in comparison to tanks.

The Glance mechanic maintains Evade as a viable defensive strategy without being so swingy as the "all or nothing" hit or miss binary we're used to from RPGs.

[Ilium]

What about introducing another variable alongside evasion and accuracy? Luck is a reoccurring theme of Final Fantasy, which could realistically factor your Hit-to-Crit ratio in a simple formula.

[kablizzy]

I'm actually a fan of luck, although it'd take a bit of work to make something with that work. I'm also not as big into Evasion formula, so I dunno. I may sit towards the back on this one.

[Mordred]

Here's a quick example to show how Evade as currently implemented (the hit-or-miss version) is borked: imagine the following characters in a battle, they are all the same level and have level-appropriate gear.

Lancer Attacker
Knight Target
Ninja Target
Wizard Attacker

Lancer Attacker (LA) has to be able to do sufficient damage to KT to make him a threat, especially since Lancers are attack-based classes. How many hits should a Lancer take to kill a Knight is a question to be left up to dev, but let's just say 10 for the sake of this example, because Knight is a tanky class. Knight is a damage sponge, so let's say KT has 500 HP. LA is dealing out 50 HP damage per attack. You want KT and NT to have roughly comparable survivability coming from different sources, so statistically, if you want NT to be able to survive 10 attacks from LA, you need to give him lower HP and higher Evade. Let's say NT has 80% total P-EV. That means NT needs 100 HP.

Lancer Attacker - deals 50 damage per successful attack
Knight Target - 0% EV, 500 HP
Ninja Target - 80% EV, 100 HP

So, this kind of works, if you're ok with combat being a huge gamble for NT. Two hits from LA and he's dead. High-risk, high-reward, sure. But we forgot someone.

Wizard Attacker

WA has to be a credible threat to KT, or else magic sucks and no one should use it. So let's say WA does the same damage as LA, 50 per hit. Except NT doesn't have 80% M-EV. Magic will crucify NT, without anything he can do about it.

If your solution to this is "just give NT 80% M-EV," then you've also screwed magic, because P-EV and M-EV being the same value makes mages relatively less powerful and versatile than fighters. Fighters have the ability to do DD without skills (base attack) plus wide selections of skills from their ability sets. Mages don't have the DD option, so if you nerf their spells/skills with a global M-EV buff, you have consigned them to uselessness. Oracles and Time Mages are especially fucked, because their effects are of the all-or-nothing type to begin with.

The hit-glance concept is what I came up with to balance dodgy vs tanky vs magic. Whether you like my specific solution or not, the problem that it seeks to solve has to be addressed.

Luck being a factor in crit chance doesn't solve the problem, though it doesn't hurt either.

[Schwerpunkt]

Luck has always just be a curve on the RNG. If it's going to be implemented as anything even vaguely like that, I don't want it. But if it's going to do something more concrete, like increase chance to crit, I'm all for it.

And Mord, my point about blocking wasn't that it'd absorb a fixed amount of damage. Rather, each shield would probably have a specific block rating (or whatever) which, as a percentage, would reduce incoming damage. So if your shield had 5 BR, you'd take 5% less damage. If your shield had 95% BR, you'd take 95% less damage. This would make it much more item-dependent than "hey, you didn't get a full-on hit, therefore you only do 50% damage regardless of whether the target is level 1 or level 50."

Other than that, I'd like to see dodge remain as a 100% damage reduction but simultaneously be a much harder stat to achieve. Like, 20% Dodge would be considered really good.

[Ilium]

For those who struggled through more of v2 than I did, tell me, was the only two evade statistics taken into consideration for accuracy P-EV and M-EV? How was Weapon Evade calculated? What I'm wanting to know was if HoI followed the traditional evasion formula from the original game.

[Ilium]

Apr 30, 2013 17:04:05 GMT -6 Mordred said:

In FFT, there's nothing stopping a Knight from having Weapon Guard, Defender, Kaiser Shield, Maximilian, Feather Mantle, and Grand Helmet all at once. So, if you get through the 100+% P-EV, the character doesn't even give a fuck because he has 500 HP.

Evasion wasn't additive in the original game, it was multiplicative. Was evasion stacked in the way you described, at HoI ? Today I've been toying around with a few different equations using the original game's formula, and using different weapon/shield/accessory combinations with Knight. Personally I haven't encountered any combinations that look broken. Of course the build in the quote above had like 10% chance to hit from the front and side, and something like 50% chance to hit from the back, but that build will probably be close to impossible to achieve in this game. And honestly if you even make it long enough through this game to get one or more of those items equipped, should they not be able to reap the benefits of doing so?

[Mordred]

@schwer: Sure, something like variable damage reduction as part of the overall "glance" mechanism sounds like it would be easy to automate. Though at that point we're kind of branching out from the original issue I was attempting to address.

As for Dodge being 100% damage reduction, note that "Miss" is still an outcome in my scheme - it's just a lot less common than it used to be. If you were to stack EV, you'd increase the chance that opponents outright miss, but not more than about 10% I'd think. Some flat-out misses/total deflections are cool, but too many makes formulas sad.

If you wanted to do something with DR working with armor/dodge, which is a pretty natural, if somewhat more involved, outgrowth of the glance-hit concept, you would effectively increase the range of damage and need to change the attack hit and damage formulae a little. What I mean by that is, in my initial scheme the basic formula for attack damage is WP*PA, then the hit/glance is handled by a random roll governed by something like this
hit% = (base hit%) * (100-EV)/100, where failing this check means you glance
glance% = (base glance%) * (150-EV)/100, where failing this check means you miss
(yes these are horrificially simplified)

To incorporate variable DR, all that would have to be done is to directly incorporate the "how good is your hit" calculation into the attack damage, so the attack formula would be WP*PA*(50-[1...EV])/50. This is kind of like how FFT handles critical hits.

@ilium: IIRC it was additive on HoI - else why would the caps be set up like they were? - but someone who was around for more of the game should chime in on that.

Regardless, no matter how you calculate your final Evade number, it's not possible to balance HP versus damage if one of the two main schools of defensive strategy revolves around traditional hit-miss evade results, as my example shows.

[kablizzy]

Evasion was additive, if I remember right, and the caps were something slapped on at the end because development never finished.

[Ilium]

Let's remove ourselves from all of this conjecture and actually work through some equations to see where we could possibly need improvement.

Physical Attack Evasion
from front:  
hit% = [(base hit%)*(100-P.CEv)*(100-P.SEv)*(100-P.AEv)*(100-WEv) / (10^8)]

from side:  
hit% = [(base hit%) * (100-P.SEv) * (100-P.AEv) * (100-W.Ev) / (10^6)]

from rear:  
hit% = [(base hit%) * (100-P.AEv) / 100]Magick Attack Evasion
any position:  hit% = [(base hit%) * (100 - M.SEv) * (100 - M.AEv) / 10000]
So far, super simple algebra. Let's introduce a Knight, an Archer, and a Black Mage for testing purposes.

Class: Knight
Level: 1
Brave: 65
Faith: 60
HP: 54
MP: 16
PA: 5
MA: 3
Evade: 0

Broadsword (WP 4, EV 5) Escutcheon (P 10, M 3)

Class: Archer
Level: 1
Brave: 60
Faith:60
HP: 49
MP: 20
PA: 5
MA: 3
Evade: 10

Bowgun (WP 3, EV 5) Escutcheon (P 10, M 3)

Class: Black Mage
Level: 1
Brave: 60
Faith: 65
HP: 31
MP: 45
PA: 3
MA: 5
Evade: 5

Rod (WP 3, EV 20)

Sword = PA * WP
Crossbow = PA * WP
Rod = PA * WP
Fire = CFa/100 * TFa/100 * MA * 14
Fire 2 = CFa/100 * TFa/100 * MA * 18
Fire 3 = CFa/100 * TFa/100 * MA * 24

Scenarios 1-4: Archer/Knight vs. Archer/Knight
front = [(100)*(100-0)*(100-10)*(100-0)*(100-0) / (10^8)] = 90
side = [(100) * (100-10) * (100-0) * (100-0) / (10^6)] = 90
back = [(100) * (100-0) / 100] = 100

Sword = 5 * 4 = 20
Crossbow = 5 * 3 = 15
Scenarios 5-6: Archer/Knight vs. Black Mage
front = [(100)*(100-5)*(100-0)*(100-0)*(100-0) / (10^8)] = 95
side = [(100) * (100-0) * (100-0) * (100-0) / (10^6)] = 90
back = [(100) * (100-0) / 100] = 100

Sword = 5 * 4 = 20
Crossbow = 5 * 3 = 15
Scenarios 7-8: Black Mage vs. Archer/Knight
front = [(100)*(100-0)*(100-10)*(100-0)*(100-0) / (10^8)] = 90
side = [(100) * (100-10) * (100-0) * (100-0) / (10^6)] = 90
back = [(100) * (100-0) / 100] = 100

any position = [(100) * (100 - 3) * (100 - 0) / 10000] = 97

Rod = 3 * 3 = 9
Fire = 65/100 * 60/100 * 5 * 14 = 27.3
Scenario 9: Black Mage vs. Black Mage
front = [(100)*(100-0)*(100-5)*(100-0)*(100-0) / (10^8)] = 95
side = [(100) * (100-0) * (100-0) * (100-0) / (10^6)] = 100
back = [(100) * (100-0) / 100] = 100

any position = [(100) * (100 - 0) * (100 - 0) / 10000] = 100

Rod = 3 * 3 = 9
Fire = 65/100 * 65/100 * 5 * 14 = 29.5

Percent of Damage Done to HP

	Knight	Archer	Black Mage
Sword	37%	41%	64%
Crossbow	28%	31%	48%
Rod	17%	18%	29%
Fire	50%	55%	87%

From this point we can argue over starting HP, PA, MA stats, and also Evasion stats for armor and WP stats for weapons in order to change chances to miss and percentages of damage done. However the equations as they are presented so far look perfectly capable of handling what we expect. Short, simplistic battles in the early game. Let's look at more advanced examples.

Class: Knight
Level: 25
Brave: 70
Faith: 60
HP: 454
MP: 41
PA: 12
MA: 6
Evade: 0

Parry

Coral Sword (WP 8, EV 5) Aegis Shield (P 10, M 50) Power Gauntlet (PA 1)

Class: Archer
Level: 25
Brave: 60
Faith:60
HP: 349
MP: 95
PA: 9
MA: 7
Evade: 10

Windslash Bow (WP 7, EV 0) Flame Shield (P 30, M 0) Mage's Cloak (P 18 M 18)

Class: Black Mage
Level: 25
Brave: 60
Faith: 70
HP: 131
MP: 370
PA: 6
MA: 11
Evade: 5

Rod (WP 3, EV 20) Elven Cloak (P 25 M 25)

Scenarios 10-11: Knight/Archer/Black Mage vs. Archer
front = [(100)*(100-10)*(100-30)*(100-18)*(100-W) / (10^8)] = 52
side = [(100) * (100-30) * (100-18) * (100-0) / (10^6)] = 57
back = [(100) * (100-18) / 100] = 82

any position = [(100) * (100 - 0) * (100 - 18) / 10000] = 82

Sword = 12 * 8 = 96
Bow = 7 * 9 = 63
Rod = 6 * 3 = 18
Fire 2 = 70/100 * 60/100 * 11 * 18 = 83
Fire 3 = 70/100 * 60/100 * 11 * 24 = 110
Scenarios 11-12: Knight/Archer/Black Mage vs. Black Mage
front = [(100)*(100-5)*(100-0)*(100-25)*(100-0) / (10^8)] = 71
side = [(100) * (100-0) * (100-25) * (100-0) / (10^6)] = 75
back = [(100) * (100-25) / 100] = 75

any position = [(100) * (100 - 0) * (100 - 25) / 10000] = 75

Sword = 12 * 8 = 96
Bow = 7 * 9 = 63
Rod = 6 * 3 = 18
Fire 2 = 70/100 * 70/100 * 11 * 18 = 97
Fire 3 = 70/100 * 70/100 * 11 * 24 = 129Scenarios 13-14: Knight/Archer/Black Mage vs. Knight
front = [(100)*(100-0)*(100-10)*(100-0)*(100-5) / (10^8)] = 85.5
side = [(100) * (100-10) * (100-0) * (100-5) / (10^6)] = 85.5
back = [(100) * (100-0) / 100] = 100

any position = [(100) * (100 - 50) * (100 - 0) / 10000] = 50

Sword = 12 * 8 = 96
Bow = 7 * 9 = 63
Rod = 6 * 3 = 18
Fire 2 = 70/100 * 60/100 * 11 * 18 = 83
Fire 3 = 70/100 * 60/100 * 11 * 24 = 110
Percent of Damage Done to HP

	Knight	Archer	Black Mage
Sword	21%	28%	73%
Bow	14%	18%	48%
Rod	3%	5%	14%
Fire 2	18%	24%	88%
Fire 3	24%	32%	98%

These are builds without any of the HP adding items like armor and such. So the HP/MP is current base using HoI growth rates. Already we are seeing a few problems with these classes. Primarily, the percentage drop off of damage done is tremendous. Not only has evasion continued to rise, but HP has grown much faster than damage. At level 25 these numbers aren't going to work, so we are going to have to look into ways of fixing this.

1. The main thing that needs adjusted is growth rates. HP per level and MP per level will need to be reduced across the board.
   
2. PA and MA growth rates seem fine, however we could try buffing them slightly going into the beta. We can always roll back later.
   
3. Items will need to have lower Evasion and HP modifiers. Evasion is not so much a problem as HP is. I didn't even have to add HP buffing items like armor to already see a drastic difference in HP to damage.
4. Some weapons could use buffs. Bows seemed a bit weak, and rods were obviously useless. We need to give mages a better WP choice than max 4.
5. Maintaining all of the above, we can come up with some better numbers that provide us with slightly higher hit rates and a higher damage-to-HP percentage, without having to include any additional formula to the calculations.
   

[Mordred]

Your math works, but I don't see the part where Evasion is accounted for in terms of its results on expected actual damage. Am I missing the point of this exercise?

[Ilium]

Why does evasion have to be accounted into damage? Evasion is a hit or miss value. If you would like to add a function for glance, it can be worked into the existing formula in a similar way critical chance already is, though I don't see the need for it. From the actual math, it is obvious what we need to work on first and foremost to better balance HP and damage.