Seyha wrote:If I can't see what the RNG seed is or when it begins or how long it lasts then what is the point of spreading the tests out? I would have to do thousands of attacks spread out over several hours to make any difference and how do you know the RNG even effects evade rates? I've never seen anything to suggest that it does.

No offense but I think you're over-complicating this.

no, it is not over-complicating things.

the RNG seed is based on the engine and libraries they are using.

and the long duration between RNG Seed Fetches is why mythics RNG has been accused of horrible streakyness. because it is subject to that due to the long fetches between RNG seeds.

RNG = Random Number Generator. they are systems/methods that deal with logically attempting to get a random #. there is no true method of getting a RNG on computers, as you must have a random source. the closest method currently of getting RNG Seeds is to capture the computers clock on a long setting to capture the most #s at a time and store that in a variable. then you attempt to pick random #s out of that value to put together your "random number".

the longer you use the same "seed" the more often you will end up with the same set of results from the same tests.

but capturing the time frequently adds a fair amount of overhead, and capturing a new seed for every "test" of that system can bog down the system when you are running something like a MMO server, so what they do is capture the "seed" on fixed intervals. the more tests you do on the same "seed", the less random and variable your tests will be.

TL;DR :

you have to separate your tests into sections across at least 15 minute breaks in between them to ensure you are getting accurate random results.

Murzin wrote:TL;DR :

you have to separate your tests into sections across at least 15 minute breaks in between them to ensure you are getting accurate random results.

If I can not see the value of the seed then I don't see how this helps us. I could literally spend six hours performing one test and the results would still be skewed by a bad RNG streak.

Look, each test took over an hour to complete. If what you say is true, then each test represents the average of 4-5 seeds.

if you are going to do 2,000 tests


you do 200 tests for each condition

then 200 more

200 more

etc

etc

as it is, there is some inconsistencies in your data

Murzin wrote:if you are going to do 2,000 tests


you do 200 tests for each condition

then 200 more

200 more

etc

etc

as it is, there is some inconsistencies in your data

That is a huge amount of work for an unknown amount of gain. Can you show statistically how this would mitigate the chance of a bad RNG streak?

I know you're trying to help Murzin but my testing is very consistent to within about 5-7%, with maybe one test being way off. If the staff take my tests and average out the results we will have a defense penetration system that is very close to live-like.

Attacker:pala 1+12 thrust , str 248, dext 302 ,WS 909 , lvl 51 weapon artefact
Defender: heretic dext 168 , ,shield 50 , WS 749

Total: 1491
Hit: 782
Block: 647
Miss: 62
Blockrate : 43.39%


Same paladin vs heretic unbufed , paladin has lvl 32 weapon 10.8 dps

Attacker: pala 1+12 thrust , str 248, dext 302 ,WS 909
Defender: heretic dext 168 , ,shield 50 , WS 749

Total :1600
Hit : 446
Block 956
Miss: 198
Blockrate 59.75%

Also with weapon 1.2 dps i got 58.59% blockrate , that means lower lvl weapon on attacker increases block for defender, and lvl 1 or lvl 32 weapon has same effect which means the lvl of the weapon is caped at a certain lvl above lvl 32 for a lvl 50 character.

PS : me and blue knew about this i forgot to post it:P.

So with the first test the pally had a level 51 weapon, and the second it had a level 32 weapon?

Yes, and i ve used a 1.2 dps weap and block was the same..seems it doesnt get higher cause of weapon lvl from a certain lvl above lvl 32 i gues.

What dps was the weap u used on your tests? ..because it affected all of em alot..and maybe some of em are eronated because of this feature.

Right now I am doing a test with the following setup:

Target B (the same Nightshade as before, no change)

The same attacker as before, but with a level 50 diamond seal weapon.

Here is the test with the level 1 weapon:

Test 2B3 - Blades 50, Strength 122, Weapon Skill 733, lvl 1. weapon
-Attacks: 1552
-Evades: 592
38.1% evade rate

Now, with the level 50 weapon and 500 attacks so far, the result is a 38.6% evade rate.

Obviously we must wait for the test to finish, but i'd bet that weapon level only comes into play with block because the shield has a weapon level too.

Which means that parry is probably effected by this same mechanic.

But not evade.

Artifacts have a hidden to hit bonus (like champion weapons, slayer and dragon weapons).

Does this affect your chance to get blocked?

Glac, if you are testing with a lower level weapon, the bonus on each must remain the same

if you are testing blockrates, the bonus on the shield and armor matters vs the bonus on the weapon

I finished my weapon level test and updated the OP with the results.

Murzin wrote:Glac, if you are testing with a lower level weapon, the bonus on each must remain the same

if you are testing blockrates, the bonus on the shield and armor matters vs the bonus on the weapon

Armours affect missrate only.

bump

I posted these tests over 3 weeks ago and they haven't received so much as a nod from the staff. That's a bit frustrating considering I spent over 25 hours conducting these tests and preparing the reports.

I hope the staff realize that without a correct defense penetration formula melee combat on Uthgard is completely broken for all classes and not live-like.

Something as simple as a "we'll look into it" would suffice.