Goal: to debunk the attribute system. It is common knowledge that attribute system gives more and more bonus damage until a certain value of attribute difference between attacker and defender is reached, at which point it keeps the limit value. The first goal of the experiment was to identify the boundary values. These were revealed to be [-220,220] points of attribute difference, giving respectively 25% damage fall-off or 25% damage boost. During the experiment, a Tyrr Duelist 110 with disabled random damage was using https://l2wiki.com/Sonic_Flash_-_Tyrr_Duelist on an Aeore Cardinal 110. Duelist was wearing a https://l2wiki.com/Helios_Dual_Sword. Cardinal was equipped with an https://l2wiki.com/Eternal_Tunic Robe Set with no set option. The experiment went the following way: a clean hit with 0 attribute difference was used to establish a base value. Next, a difference of -280 attribute was set between the attacker and defender characters. The attribute of attacker was increased by 10 each step, and damage was noted down for each value. The raw data referred to the base value can be seen on the plot below: The next goal of the experiment was to identify the function above, to be able to write it down in it’s analytic form. Details on obtaining the analytic form can be found inside the Math Corner: Spoiler: Math Corner Let x be the difference between attacker’s and defender’s attribute values, and f(x) the attribute formula. It can be immediately observed that f(x) is a discontinuous function. It can also be deduced that f(x) will take form of where sat is a saturation function defined as y(x) is an internal function, and lb and ub are the lower and upper bounds. We should gather knowledge on y(x) before we find a suitable candidate. It should be noted that the y(x) function has several properties similar to the signum function. An assumption was made that where and P(x) is a function that we will attempt to identify numerically. Due to the shape of the results, it was assumed that P(x) is a polynomial function. Starting from the reasonably simplest polynomial of degree 2, an algorithm using least-squares criterion and Vandermonde matrix was coded to find the coefficients. It was also guessed that P(x) takes absolute value of X as an argument. The algorithm gave us the following function: After plugging it into signum and sat functions, we have received the following fit: The result is pretty much close to perfection, making it unreasonable to check higher degrees of polynomials. All that was left to do was simplification. Observation: 0.0191 is related to a famous mathematical constant: Result: the attribute system factor used in damage calculation (base damage is multiplied by the factor) is calculated using function: where x is the attribute difference between attacker and defender. The function plot can be seen below: The maximum damage bonus/fall-off is +-25% at +-220 attribute points difference. It has also been tested that Dragon Weapons Stage 1-3 do not break this cap, like they do with Maximum HP.
Ah, a working tester! We are grateful! Read the missing requests and lets see. Or make new ones if you wish (base rate of tests per week? ) )
Rate - hard to say, depends on how complex a test is, at least 1 per week but I hope for 2. I've got a (secret) list of subjects I'd like to work on first, because they cover core systems of the game we do not understand yet. After that I can move on to less-critical requests.
Weird, I remember it was 220 + /- and 25% dmg difference, but I had the impression it had small breakpoints inside the 220, like 100 or 120, 20 and 50, that it did small jumps and after it was stable for a little while.. pbb mistake on my part Seem fun having your test playground. Dunno if you request on that other topic or here, but if you could test dmg formulas and crit reduction from dagger (since I remember drum said -30% crit reduction, but it only reduced around 2-4% crit dmg) and how exactly dmg reduction works (and if there is difference between pvp reduction and dmg reduction, how they multiply each other or not), should be nice..
Thats why the retest, attribute system was changed multiple times. I've also tested 220 difference vs 1k difference (dragon weapons) and damage was the same
well, as for me, and i think most of other players who is interested in this stuff it's still too hard to understand differences. better would be: 0 atr - 1000 dmg 50 atr - 1050 dmg 100 atr 1100 dmg 300 atr 1300 dmg it's just an example. or even better would be if you test it on specific mobs how attribute attack increase damage, not some tables and formulas (which can call a satan if you read it wrong), but exact numbers.
I can't, because it doesn't work that way, which was the point of the experiment. All that matters is your attribute attack value, minus enemy's defense to that attribute. If you have 820 attribute and he has 600 defense, congrats, you get 25% bonus damage (1.25 times base damage). If you both have 820, you don't get any bonus. If he has more defense (820) than your attack (600), that's -220 attribute -> 25% penalty, meaning you deal 0.75 times base damage. If you try to trick the game by having 0 attribute attack (so your character sheet says "None: 0"), the game will automatically calculate damage versus enemy's Holy attribute (your 0 vs enemy a lot, so 25% damage penalty) What can be more exact than the last plot? 1. Take your base damage - base means you have same attribute attack as your enemy's defense. 2. Get the actual value of your attack attribute, minus enemy defense (which you need to know, but in PvP you can very well theorize what are the ranges of enemy resistances). So you have 800 attribute, enemy has 650 defense. That's 150 difference. 3. Check the last plot for 150 on x, says 1.16, means you have 16% bonus damage. 4. Multiply your base damage from #1 by 1.16 -> that's your final damage. If you have less attack than enemy's defense, scenario is the same, but for example 600 attack vs 700 defense -> difference is -100 (minus). From the last plot you read that -100 attribute is 0.91 (9% penalty). You take the base damage, multiply by 0.91, and that's your final damage. You can also get base damage value by having very low attribute attack versus high defense enemy, and dividing it by 0.75. As for the mobs: each mob in the game has different resistances, set arbitrarily by the developers. What's more, each mob can have totally different values of different attributes: much more fire, much less wind, etc. Checking even the 101+ spots only for each of the 6 attributes on every mob would take weeks, which could be spent doing more beneficial experiments. There's a general rule of thumb in terms of attributes in PvE though. Even when a mob does not have "weakness to certain attribute", as long as it has one of the following "races/types": Humanoids have increased earth resistance and decreased wind resistance. Demons have increased dark resistance and decreased holy resistance. Undead have decreased holy resistance. Angels have increased holy resistance and decreased dark resistance. The differences won't be higher than 50 though. Due to the fact that on first Goddess of Destruction most of the enemies were either Undead or Demons, it was common sense to use Holy as the PvE attribute. Now it mostly depends on what your daily targets are, but Holy's still a good choice, because there are almost no Angels on the endgame.
Depends on the spot, mob type, and which attribute you use. You need 220 more than your enemy. Last time I checked Elven Village, mobs had around 550 defense in holy, and 600 in earth/wind (depending on a particular mob). Meaning you'd need 770-820 attack. But older spots like GC and Enchanted Valley had less, like 500. Don't know about new Sea of Spores.
Wow i realy thought that i would never see a decent and interesting experiment GJ dude and keep up the good work
Quick question. When mob has an icon, i.e. "Earth attack: strong against element of earth", does it mean it deals damage with earth attribute but is weak to earth attribute itself? Or both it deals earth and is strong against earth (weak against wind)?
