Whats better?

[Tsuko_Asura]

Unless its the difference between reaching or not reaching a crit. tier, Toreador's.

[Amele]

Unless its the difference between reaching or not reaching a crit. tier, Toreador's.

^ this, but you may find Iota easier to get than toreador's if you have neither and aren't planning to do a mythic weapon.

[ringthree]

Unless its the difference between reaching or not reaching a crit. tier, Toreador's.

From the evidence I have seen I believe that Crit hit functions like the rest of the stats in the game, the more Dex you have the greater the effect. I do not believe there are tiers at all, but rather limited initial tests that were done at different DEX levels that gave that impression because of the curve of the function.

For this reason I would suggest against worrying about DEX unless you already have a lot of DEX in your build. I had a graph of someone else's much more comprehensive tests around here somewhere...

[Amele]

From the evidence I have seen I believe that Crit hit functions like the rest of the stats in the game, the more Dex you have the greater the effect. I do not believe there are tiers at all, but rather limited initial tests that were done at different DEX levels that gave that impression because of the curve of the function.

For this reason I would suggest against worrying about DEX unless you already have a lot of DEX in your build. I had a graph of someone else's much more comprehensive tests around here somewhere...

searain's tests showed a definite tiering* to the function. google search for something to the effect of seaseaseasea livejournal ought to find it.

this would not be the first thing to operate in tiers (see: resist rates, pdif, etc)


*ok, technically it was a largely linear function with stepping and not strictly speaking flat tiers; but since the person you're quoting was using tier in its general purpose sense this is a minor distinction

[Max™]

The tests showed a smooth, and slow, increase until roughly a multiple of 10 Dex over target Agi, when you got a jump larger than the previous +1 Dex changes had shown.

This repeated over +10, +20, +30, +40, and +50 Dex, and over +50 Dex apparently it showed no further increase in crit rate from Dex, suggesting either hugely diminished return rate, or a cap.

Thus, tiers.

It's worth using to aim for the closest reasonable tier you can hit, and it explains why older tests showed one person saying "zomg, Haidate give +X% crit rate", and another gave different results.

[Meat]

^ this, but you may find Iota easier to get than toreador's if you have neither and aren't planning to do a mythic weapon.

I've got 2 toreadors and 40k ampules, I'm jut wondering if there's even a point in keeping the toreadors. I'm not looking for an increase in acc just looking to avoid a decrease especially when I can bank about 6 mil for another salvage body piece.

[Amele]

I've got 2 toreadors and 40k ampules, I'm jut wondering if there's even a point in keeping the toreadors. I'm not looking for an increase in acc just looking to avoid a decrease especially when I can bank about 6 mil for another salvage body piece.

if your dex is 1 greater than a multiple of 4 (ex: 65, 69, etc) in your normal gear and food, excluding Iota ring, a 2 handed job will get 6 accuracy out of the ring. if your dex isn't 1 greater than a multiple of 4 you will get 5 accuracy out of the ring.

so I guess the question is, is it worth 1-2 accuracy to you to gain 6 mil?

[FailureMidgard]

if your dex is 1 greater than a multiple of 4 (ex: 65, 69, etc) in your normal gear and food, excluding Iota ring, a 2 handed job will get 6 accuracy out of the ring. if your dex isn't 1 greater than a multiple of 4 you will get 5 accuracy out of the ring.

so I guess the question is, is it worth 1-2 accuracy to you to gain 6 mil?

You also gain a decent defensive-ish ring with Iota!

[Fosnofx]

Hey,

I am aware that this topic has been brought up but can't seem to find a definitive answer.

Mercurial Kris DRK Zerg.

HP (Intruder's/Morukaka's) or Wyvern Earring?

Thank you.

[Meat]

haste

[Sanoske]

70 MNK. Cobra Unit Leggings(Acc +5, Store TP +3, Enmity -4) or Strike Subligar(Kick Attacks +3, Eva +3)?

Would the Acc and small amount of Store TP make them better for TPing in until I get DLegs? Not a WS piece for me since I use Kensei Sitabaki for the VIT mod atm.

Just for reference, I'm not leveling past 72 until I get Destroyers/DLegs/DHead/DBody, and won't be hitting 75 until I get a pair of Haidate.

Also, the head piece in a see a lot of MNK WS setups on DK and HF, is that Genbu Kabuto or Shura Head? I know VIT is a equal/larger mod in both WS, so I assumed it was Genbu(since it would be VIT +15 vs STR +5/Acc +5), but I wasn't sure.

[SamanosukeShiva]

cobra legging is better

i personally use genbu for HF/DK

[Phalanx]

For melee zerging for example KV,Bv2, and Kirin; which one is better relic scythe or KC?

[ringthree]

searain's tests showed a definite tiering* to the function. google search for something to the effect of seaseaseasea livejournal ought to find it.

this would not be the first thing to operate in tiers (see: resist rates, pdif, etc)


*ok, technically it was a largely linear function with stepping and not strictly speaking flat tiers; but since the person you're quoting was using tier in its general purpose sense this is a minor distinction

Right, kind of. LOL It is probably three linear functions, not really "tiers" as they were explaining it, and I think that it is more than a minor distinction. It's like what Ruk suggested that it was with accuracy, except kind of in reverse, where more accuracy will have less of an effect the closer you come to the cap, while DEX appears to have more of an effect on crits as you widen the gap between your DEX and the mobs AGI.

I tossed these together to better explain the situation as I see it.

For accuracy: http://i236.photobucket.com/albums/f...racy_graph.jpg

but for Crits: http://i236.photobucket.com/albums/f...Crit_graph.jpg

So there are "tiers" in the idea that from what we can see there appear to be three different functions for most calculations in the game that check against mobs stats. (Player stat is below target stat, player stat is equal to target stat, and player stat is above target stat). But I think people should move away from saying there are "tiers" and rather look at it like a curve, where stacking stats is going to be better than just tossing on some random amount of a stat to hope to improve the result.

I would not be surprised at all if all game combat calculations worked in the same manner. It would explain elemental resists also.

[ringthree]

For melee zerging for example KV,Bv2, and Kirin; which one is better relic scythe or KC?

As of now, I believe it has been theoretically proven to be KC.

[Tsuko_Asura]

Indeed. The best option, however, is to use Catastrophe, and then immediately swap to KC.

[Amele]

you and I have had similar discussions before I think.

I don't know if I agree with ruke yet on the accuracy thing but I have to admit that I don't have time to do extensive testing myself. I personally felt naga did an excellent job in the pre-90% range, but I agree that it could bear more testing in the 93-95% range.

that said, I don't believe that SE would have made the accuracy calculation that complex. (the pdif calculations aren't really that complex either except for the 'sticky' range around 1.0 which appears to have been a design goal, rather than a design consequence.)

I should do a similar graphing to show what I believe the crit formulas to be but basically, imagine a slightly sloped stairstep and you'll have it.

I figure there's about 5 steps. (the formula itself wouldn't be complex, it's basically two terms: one that varies based on dex-vit and another that varies with a larger coefficient based on (dex-vit)/C where C is probably 10).

[ringthree]

The tests showed a smooth, and slow, increase until roughly a multiple of 10 Dex over target Agi, when you got a jump larger than the previous +1 Dex changes had shown.

This repeated over +10, +20, +30, +40, and +50 Dex, and over +50 Dex apparently it showed no further increase in crit rate from Dex, suggesting either hugely diminished return rate, or a cap.

Thus, tiers.

It's worth using to aim for the closest reasonable tier you can hit, and it explains why older tests showed one person saying "zomg, Haidate give +X% crit rate", and another gave different results.

Actually, if you are talking about Searain's tests then I completely disagree. Those tests showed a rather smooth and correlative increase of crits to DEX. I have graphed it in XL before, but can't find it now, so give me a minute and I will do it again.

Edit:
In further reply, the original testing of this that created the notion of "tiers" of crits wasn't very accurate, and I believe only tested a few different amounts of DEX and the Crit results. It did show there was an obvious correlation, but the nature of the correlation was unclear. Searain's testing, was definitely better, but still has flaws in that there were gaps in the amounts of DEX and in a perfect world we would want multiple tests on each amount of DEX.

It should be noted that the best tests (Searain's) only had a range of 28, so I don't know where you got the information about the different tiers of results from tiers of DEX, and if they are from someone else's tests I would love to take a look at the numbers!

As for Searain's test, the one that I believe to be the best so far, here are the results. http://i236.photobucket.com/albums/f...p/GC_Crits.jpg

Honestly, I do not see how "tiers" can be drawn out the information, as there are several places that have significant gains in crits from increased DEX that can't either be attributed to either a raw or a comparative increase of 10 DEX.

Obviously, as always we need a lot more testing.

Edit 2: Here are some theoretical ideas for the functions as both Amele and I see them.
http://i236.photobucket.com/albums/f...t_theories.jpg

[Callisto]

RNG Bolts for TP/WS? I have acid/holy/venom/darksteel to choose from, no dedicated holy bolt build though. Acid for TP, Darksteel for WS?

[Amele]

there's an absurd jump from 108-110

so ok, his tests really only show one stepwise. (I said linear functions with steps) I'll do a plot sometime and stick it on my photobucket to show you what I mean.

basically it's 'several linear functions' similar to your supposition, but with less variation in slope between the functions and more variation in their +constants.