A lot of games use 1DN rolls, they are quick, easy to implement, and WRONG.Anybody played . . . any tabletop game, or sit down D&D ever, how many dice did you bring to the table . . . I'm going to guess more then 1. (like a small bag of them). Just one action on a tabletop game will take upwards of 5 dice easy. Why? Because multiple dice = somewhat predictable behavior.
You're standard 1DN takes 2 operations, + Random # generation or lookup. The generation of Gaussian Random numbers adds an extra 5 or 10 operations, more if you use a low efficiency algorithm, and then takes get this 2 operations. If you do your RNG in advance or during slow cycles with not much on the CPU. it's the same!!!So you've got a epic dude of Epic-ness kickass sword, expert training, Years experience. This guy is made of win.
This guy encounters a spider in the woods. Attacks, Rolls for 0 - 85 attack. gets a 3. doesn't kill the spider. Wait, What??
I understand there maybe should be some chance to miss. but then just say miss, don't embarrass . . . everybody . . . with a hit of 3.
If your TOP hit is 85. you aught be doing 60 - 65, reliably.
that's 20 + 17D5. which is a lot of operations, or Gaussian(62.5, 5) which is 2, and a Gaussian look-up.
Same goes for Armour, You're amour that gives 8 protection should almost never roll 0, unless the attacker has some special for a chance to ignore Armour.
1DN should be used exclusively when it is logically defensible that all possible outcomes aught be equally probable . . . which is in thermodynamics, or nearly never.
If anybody thinks 1DN rolls are a good idea, Speak now, or rally behind me in this crusade against the misuse of random damage, and all other occurrences of 1DN.
Robbie Price
Would something like this work?
If Attack Value was 20, Level 2 Creature
5+((Random Number (2*2) to 15))
In this case 5+(Roll of 4 to 15 or Min 9 Max 20
25% of Attack Value + ((Random Range(Level*2) to 75% of Attack Value)) Note May or May not want to MAX the Level, needs testing.
I am Adding Level as most of the Attack Value cones form the Weapon, and one would think the more Skilled would do a bit more damage with the same weapon as a Trained new Fighter.This gives a large Range and with DEF working the same way still allows Chances for Misses when the Values of ATK and DEF are close. But also allows for a Strong Hit as well.
I was thinking also Squads could use someithing like this:
Three Warrior Squad Attack Value 121st Attack at 122nd Attack at 143rd Attack at 16
Adding 2 to the Value of Each Attack after the first, As the Warriors are Trained to work as a group.
Each then would use their Attack Value vs the DEF, so if the DEF was say 20 on the above example, the 1st and 2nd would likely Miss, where the3rd would have a chance to do some Damage.
Right now I think Squards only Take the DEF Into Account once and Add all the Damage, to get the really High Damages I have seen in Game.Well just some thoughts on what may work, and still keep the rolls simple.
Lee
There was some talk about maybe fixing it for spells. But your right I've seen no response for fixing it for the melee or archery. I suppose we'll find out mid next week when the Dev's are back in the lab, or maybe they will continue to happen to not see these requests.Robbie Price
Please god make it happen
Actually, in a lot of games they use 1dN + X because it is easy for people to understand. Warcraft 3 Actually uses this method, I believe.
Agreed, but the Ranges need to be set better than 0-100, I think to have any real meaning.
As an aside, I don't think game companies give folks enough credit, not to mention some one almost always answers how thing works on the boards. The SD folks are generally pretty good about his. I think they could make a much more rich and indepth system without too much trouble or folks having a hard time understanding it. This is especially true if they'd just make a sticky on how it worked since the manual is done.
THANK YOU OH PLEASE GOD FIX THIS!!!!
This is definitely one of my primary gripes (the big three) with the current battle system.
1. (what this thread is about)
2. attack/ defense stacking, (units acting like mutant amoebas)
3. lack of abilities (direct combat, leadership, regenerative, etc)
obviously there are other complaints, like no soldier death/ reinforcement system (for large units), primitive unit placement, empty battle-maps (z values would be nice), and the list goes on.
//
Edit: Ways to Fix it (only one listed)
simplest way (maybe?) is to have your raw shit score (1DN + X) as an "attack value" vs their (1DN + X) defense value.
Then, have small chances for things like "parry", and "dodge" and document them as "Miss" in the AI (or simply what they actually are).
Then, if the attack is equal or greater than defense (or just greater) then you have a damage roll that is (Xa - Xb) say 40-65. (if the attack is lower, just say "defended")
Then, if the Defender/Target has some level of damage absorption, say 0-5 (normal) or higher for bosses and heroes ... damage is equal to damage roll - absorption roll. If absorption roll is equal or higher than damage, then say "absorbed."
And so there it is ... "parry", "dodge", (maybe miss), "defended", and "absorbed" are the only non-damage choices ... (absorbed = 0-damage, very rare)
As for actual hits ... you have your reliable damage range (the actual roll), and you ALSO have (perhaps) a critical hit chance ... maybe in this game critical could be 50% more damage, and "super critical" could be double damage. Something simple like that.
as for other ways ... just have 5 D10s + 20 or something (with an actual miss chance), a lot like other ppl are suggesting.
They're absolutely a good idea. The problem is you're basing 'how it should be' in aesthetics.
Your desire for more predictability would accomplish just that - make things more predictable. In some circumstances, that means you have no chance of losing, and in other circumstances, it means no chance of winning. To me, that's incredibly boring.
They key thing which you accomplish by generating a number 0...N is that, regardless of how much of a difference there is between the stats, there's always a chance of success or failure on an attack. That means there's no point at which the outcome becomes certain, and there's always the possibility for lower-quality units to damage a high-quality unit. In a game like Elemental, where there's open-ended tech research, squads, legendary equipment, and random monsters roaming the map, this is essential. It ensures that you'll never have a situation where one unit is completely unable to damage another - they simply have an ever more faint chance of doing so.
Having done a whole lot of statistical calculations for games I play, 0..N is actually one of the best game mechanics if you want an actual balanced system.
Now... there's a serious question of whether the game is actually doing that, because in my experience the random numbers have been dramatically weighted towards the low end of the scale. If I have an attack rating of 100, then I should have a 1-in-20 chance of doing less than 5 damage, yet in my experience I've practically never seen more than that.
It would also be nice if the game stats showed you the character's average damage, rather than the maximum. That would be a simple matter of generating a number 0..2N rather than 0..N.
Oh, and I wouldn't mind seeing something like a calculation to hit or miss, with a fixed damage rating... but I'd still want the attack/defense values to be based on the same scale.
0-N Rolls are pure, unadulterated BS. Absolute garbage, they take the Random Number Gods and elevate them to the level of Almighty. Outrageous outcomes occur way too often. I just had a sovereign, familiar, fire giant, and two squads of HP 20, Attack 20, Def 12, Mov 2 utterly wiped by one enemy squad of HP 20/24, Atk 24, Def 12, Mov 2 soldiers. I cast multiple lightning strikes from an int 17 caster (with an air shard) that did f' all damage, my 20 attack squads did f' all damage, my familiar did f' all damage, the only thing that did any damage at all was the fire giant, and half of that was because it has unstoppable damage.
Outcomes like this should be obscenely uncommon (like winning the lottery jackpot), not frequent occurrences (like winning $2 in the lottery). All I can say is that in this game at least the RNGs aren't as utterly blasphemous as GalCiv2, where they elevated quantity (of utterly garbage ships), over quality (of Super Dreadnoughts) to ridiculous heights. The RNG uber-power also exacerbates the obscene healing rates in this game, if your 200 HP troops are down to 50 HP, you may as well abandon them, because it will be a geologic aeon before they are healed and ready to rock again.
0-N HAS TO GO STARDOCK, IT'S ALREADY RUINED SEVERAL OF YOUR GAMES, DON'T LET IT RUIN THIS ONE TOO
I'd actually be more in favor of a higher base and a smaller random number. Base attack of like 10+1d4, rather than 2+3d4. Probabiliity to hit is a separate matter, and should be calculated with a separate stat. A dagger should have a high attack (high probability to hit) with a low damage. They ought to be 2 separate numbers, not rolled into a single attack value.
Goodmorning all
I think you misunderstood what I am going for, by using a Gaussian you can control very precisely the amount of randomness you have in the game. If you use narrow Gaussian you get highly predictable behavior, if you use wide Gaussians you get random behavior. (So much so that if you use a gaussian wider then the acceptance window (IE roll 10 +/- 100, but re-roll if not between 0 and 20, you get a uniform distribution back.) If for whatever reason you believe that in a particular situation a uniform distribution is best you can recreate it easily. The point is that Gaussians are easy to implement, and much more powerful, and fairly easy to imagine for most people 'almost always 10 or very close to it'. In terms of proccessing time they are not considerably slower then 1DN's and are often faster then MDN's.
If well implemented there is no reason why Gaussian based hitting, damaging, dodging and defending need be unduly predictable. All I'm saying is that the straight 1DN is a blunt tool, a unduly unpredictable tool, and the wrong tool. as such it should be used sparingly or never.
Winnihym take a look at my suggestion for separating hit from damage, would that fit your requirements? https://forums.elementalgame.com/394023 . Also 10 + 1D4 = Gauss(12.5, ___ ) where if ___ > 2.5 you put in hard barriers and then almost recieve the uniform distribution or you can have ___< 2.5 and have small but non zero chances of also rolling a very (un)lucky 9 or 15. As is, it would be approximately = Gauss(12.5, 1.15 )for interest 2+3D4 ~= Gauss(9.5, ~0.6666) where the probably of rolling outside the range is smaller then .5%. Just saying, For the right job use the right tool. Robbie Price.
The problem with this is that once your X because larger than their 1DN + X, then you are effectively immune to damage. There is no chance that you can lose. Drama and excitement are lost. Outcomes become inevitable. Can a lone knight in platemail defeat a 100 peasants? What about a thousand?
I would probably prefer the system that they have in place over that. The middle ground is taken by tweaking the distribution of the RNG. You can do something as complex as a Guassian, as Mr. Price suggests, or just average a few rolls together. Crazy stuff can still happen, but it will be rare.
That way, when the lowly peasant stands tall and defeats the infamous Black Knight, they will be talking about it for ages. Ballads are sung, heroes are made. All is right in the world, because once in awhile, despite overwhelming odds, the small man is victorious.
I think you largely missed my point. Attack and Defense should (imho) be separate from damage.
Attack and Defense can both be 0-100 for all I care (but more likely 1-30 or 5-80)
Then, damage can be more normalized (5-15 Gaussian) or (20-50 Gaussian) ... while resistance scores (if any) should be low and random (like 0-5)
...
basically, there is no reason why Xb EVER has to be higher than [Xa +1DNa]max (and that's also not even where I was going with it)
In fact, if Attack and Defense end up being (Xa + 1DNa) vs (Xb +1DNb), then we should make sure that (on raw attack/defense progression) any Xn +1DNn, at max roll, will be higher than ANY possible X. That would be a simple design philosophy so that there is never a 100% chance to miss.
HOWEVER, even if that peasant hits, its going to be doing just a little bit of damage, and if the hero has a lot of HP, then well yea ... lots of peasants WILL die whether they are lucky or not. But yes, you will be satisfied because its possible for an unlimited number of peasants to kill anything.
and yes, if infinite number of peasants aren't allowed to kill everything else, something is wrong (unless its a god)
It's not only about ranges, it's about distribution.
1DX + N is still a uniform distribution. And that is bad, because you have the same chance to get highest damage, midlle damage, lowest damage. Low and high damage should be fairly low.
When you have an attack of 100, you should estimate your rough damage output to 40~60. You can still do 1 or 100 damage, but it won't be 1 chance on 100.
I also agree with Vieuxchat that if we ARE to only focus on ONE roll (which I think is a bad idea in itself), it would be FAR better served as a bell curve. So, for instance, 66% chance to get 40-60 damage, and a 90% chance to get 30-70 damage, and a 95% chance to get 20-80 damage, and a 98% chance to get 10-90 damage, and a 100% chance to get 1-100 damage.
But! if we actually SEPARATE attack from damage, we COULD have 1DN attack vs 1DN defense, and then have a largely fixed amount of damage for successful attacks. That "may" be simpler. Just something for stardock to think about.
"Say 1-100 attack vs 1-80 defense, yet all successful attacks give 20-30 damage"
The current system is fine, but not correctly tuned. The attack is 1dN, the defense is 1dM.
In effect, this is like having attack beeing two dice and defense constant, which makes for a nice triangle curve when M and N are equal. The problem appears when N and M are widely different, such as attack 150 and defense 20. In that case, the triangle curve becomes almost a flat line...
I'll be shortly posting a new thread for combat improvments.
Also note that people asking for a real bell curve are absolutely not aware of the consequences. They would become shocked that a unit with an attack rating of, say 6 would not be able to scratch a defense 8 unit, and would thus vocalize strongly against the bell curve...
Yves
only if its a very narrow Gaussian/bell-curve.
Honestly though ... I'd much rather have 1- Na attack vs 1- Nb defense WITH!!!! a constant damage rating for successful hits. Say, a successful attack will ALWAYS do 10 damage (or 5-15 damage, or Gaussian 10 with a Standard Deviation of 2)
That is to say ... no "3 damage" and "1000 damage", but instead "miss", "miss", "oh look 10 damage", "miss", "oh look 10 damage", sorta deal. (as an extreme example)
another example would be "miss", 5 damage, "miss", 10 damage, 7 damage, "miss", 12 damage, "miss", "miss", "miss", 15 damage, 9 damage
If none of my ideas are accepted ... just go with the MOM combat system. A far superior system to what is currently in place.
What about linking the number of dice to attacker level? So a level 1 character gets 1d20, but a level 10 character gets 10d2. That would reflect a noob with a mace could come up with anything, but once you've mastered combat, your results become fairly predictable.
Heh, honestly I rly like the Idea of Champions having more dice with higher level.
(lv 1 having 1D20 and level 10 having 10D2)
of course, it'd probably need to be something a little more abstract/ flexible for decimals ... in order to work for all levels.
Maybe depending on the weapon depends on how fast a Champion hits the "invisible ceiling" for max number of dice allowed for that weapon?
As a contrast Wesnoth has a number of attacks, each with fixed damage (modified by resistances). The chance to hit is based on the defence of the unit targeted (which isa combination of the unit type and terrain)
so an elf may have 4 hits doing 6 damage with a sword. If he targets a skeleton on open ground he may have a 20% resistance, and 40% defense. This then gives him 4 attcks with a chance to hit of 60% each doing 6*(100-20%) = 4.8 which would round to 5
D&D used defence to affect the chance of getting hit, and used a 1dn damage roll
Runequest (and car wars) used a to hit roll based on attackers skill, with the armour being subtracted from the damage.
If I understand correctly Elemental uses a 1dn minus 1dm method. This is equivalent to 1dn +1 dm -(m) which already produces a non uniform distribution. (ie if attack and defense are the same and equal to 10 the roll is 2d10-10 if you do the math - this may be a point or 2 off depending on whether elemental and your definition of 1dn starts from zero or not)
This isn't too bad - it may be better not to increase the attack strength but to increase the number of attacks (but not movement) for experience (think Matrix/kungfu masters) - or give a chance for multiple strikes
Good discussion. Just a note[on-topic]: Squads/Group of units -> I think that each and every unit in a squad should have a separated attack roll / defense roll. 1 att/def roll / squad makes no sense imo.
This is one of the many improvements I suggest in this thread.
Bad idea. A level 10 has fewer chance to get a 20
(10d2 has fewer chance to get 20 than 1D20)
I'd just like to say, although I don't like the "one shots" my champions sometimes get, I REALLY like that you can "Miss" in this game. Really. I've always hated TBG that always hit; and always do damage. I love when my champions are getting attacked, and it's a close battle, and SWISH, the enemy misses
I also hate how every 0 damage attack is represented as a miss... Twelve archers firing at a dragon at close range should never miss. Do no damage probably but not miss!
There are many great features available to you once you register, including:
Sign in or Create Account
