That doesn't make sense. Your attack moves use hated. Your defense moves use discipline. Focus on discipline replenishment, so you can pop Preperation whenever you need to. Now you can spam offensive and defensive skills all you like, while not requiring any "Hate Generators". I believe that was the original intention of this post.
We're trying to lessen Discipline consumption whilst being able to maintain the "spam offensive" idea, this leaves us more breathing room when we want to use other Discipline abilities - without being tied down to X number of Multishots per cycle. Though admittedly it's the simplest solution. My "guide" post was to look at the possibilities we had... Though unfortunately it needs revising, which I think I'll do now.
Edit: And as azure says, we want to optimise how much hatred we actually generate, it's kind of min/maxing within the build.
P.S. I'm not doing this so I can pick which build style is the best dps, I want to play the most fun one... The more options there are the better the chance I'll come up with something great!
I've done some more accurate math in regards to expenditure vs. generation for both H and D. I now have some ballpark figures for what total regen values we should be aiming for to sustain and also figures for how we could achieve them. Will post them here and also add to guide shortly.
Abilities are assumed to be used exactly on cooldown.
No Passives etc are included in calculations unless otherwise specified.
D regeneration is assumed to be base, i.e. One.
This formula is used for H Generators marked with an * :
H/s = V(Dr/Dc) + B where V is the amount of Hatred the skill grants , Dr is the Discipline Regen, Dc is the Discipline cost of the skill. B is the Sum of True Passive Regens (i.e "Base" regen + Gear).
1) Prep ; 10.75H/s with No Veng. 12H/s with Veng. 13.9H/s with Veng & Perf.
2) Gear (Pure) - DualWield + Cloak + Base = 14.5H/s. 2H/Quiver + Cloak = 11.2 H/s
3) Smoke Screen ; 7H/s no Perf. 7.8H/s with Perf.
4) Shadow Power ; 6.75H/s no Perf. 7.3 with Perf.
5) Companion ; 3H/s + 4.5H/s = 7.5H/s.
Rank with no Passives: DW Pure Gear > Prep > 2H Pure Gear > Comp > SS > SP.
It is important to note that to calculate the total H generation if multiple skills are used then you must account for the fact that base regen is included in each one.
"Passive" Discipline Regeneration of each Skill
These equations/values are simple estimate as to what D regeneration we will get depending on certain gear factors and assumptions. The values will be used later on after evaluating Discipline consumption for a particular build.
1) Multishot - If we assume 5 targets are hit per use, then this gives us 5D. Assuming 1 Attack per Second, that's 5D/s. BUT, we aren't going to be spamming MS constantly - it's better to evaulate the average D/s generated over a certain time. the equation for this is:
D/s = 5N/T + B where is the Number of MS uses, T is a given time and B is the base D regen of 1/s.
For example, if we assume 3 MS uses per 15 seconds then that gives us a total regen of about 2D/s.
2) Nightstalker - this is very difficult to accurately quantify because it depends on several variables. I worked out an equation that provides a rough estimate of D/s, such that if we gain more information it will become useful. As is, it is a vague estimate based on educated guesswork. It is:
D/s = 2CP(APSbar) C is Crit chance and P is Proc Chance, expressed as decimals ideally. APSbar is the Average Attacks per Second.
So, if we had 35% Crit, the Proc chance of NS is 50% and we average 5APS then D/s is 2(0.35)(0.5)(5) = 1.75.
3) Prep (unruned) - 30D/45s = 0.66D/s
4) Prep (Focused Mind) - 45D/45s = 1D/s (I believe Cooldown remains), providing you don't cap D whilst Prep is active.
Naturally you can't use Prep for D regen if you're using it for H regen, but if you do use it for D regen you should rune it unless you prefer the instant effect of the unruned version.
Discipline Spenders - Estimations!
I've estimated how often a skill might get used if it has no duration, if it has a duration I have assumed it's used back-to-back. Sentry & Companion have been excluded, Sentry because I'm not sure if you can have multiple ones down, Companion because it's one use only.
Using these figures I've estimated rough discipline consumptions for certain build styles:
Prep Build - 3.5D/s -> 4D/s
Crit Build - 4D/s -> 5D/s
Risky Alt Build - 3D/s -> 4D/s
The riskier (the less discipline skills you have) the less D you will consume per sec on average - obviously.
Hatred Spenders - Estimations
Both Cluster Arrow and Impale have been excluded from my estimations because I believe they are both special cases that require analysis within a particular build.
Assuming 1 APS and spamming ability, or use on cooldown:
RoV - 40H (30s CD) - 1.33H/s (this must be added on top of any ability you expect to be using during the CD).
MS - 15H/s
EA - 10H/s
FoK - 20H (10s CD) - 2H/s (same rule as RoV)
Chak - 10H/s
Strafe - 15H/s?
RF - 20H -> 10H, avg ~15H/s
Basically we see that all the above abilities should be sustainable with Hatred Generation of about 15H/s - less if using lower cost abilities, or using certain gear.
Now we have rough estimates for a particular build, or what sort of generation we need - we also have estimates of how much regeneration a given ability provides.
Yes improving discipline returns can help, but you can only use hatred so fast. If we assume you have an attack speed of 1.3 you will end up using 260 hatred every 20 sec. I only use 20 sec because that is the length of time to get the discipline to use Prep. Over this 20 sec we can assume that you have a natural regen of about 10 based on gear. Prep generates 125 hatred. Overall you generate 325 hatred and only spend 260 of it. This is why I say Prep is over kill. Trying to get even more regen for either resource at this point is pointless. You have no way to burn off the extra hatred unless you are in a special case where spamming something like Strafe or Cluster Arrow is a good idea.
If for example you want to get more damage out you can switch out Prep for the Companion(Bat) you will generate less hatred but also have more damage output on top of freeing up discipline to use on other things like Vault of Sentry.
That's all well and good if all you want to do is spam an attack skill, but what if you want to do other things like Vault around a lot, drops lots of Caltrops, go invis all the time with Smoke Screen, or try to keep Shadow Power up as much as possible for 20% leach and 65% damage reduction? Even if you can get enough passive hatred regen and a slow enough attack speed to not need Prep, you are still limited by your natural Discipline regen. Multishot lets you get so much more Discipline than you could get any other way. Perhaps this is a different build--max Discipline defensive/utility that just happens to not require hatred generators, instead of a no hatred generator build that just happens to generate a ton of Discipline for defensive/utility use--but either way, it seems like a great way to build the DH.
Yes improving discipline returns can help, but you can only use hatred so fast. If we assume you have an attack speed of 1.3 you will end up using 260 hatred every 20 sec. I only use 20 sec because that is the length of time to get the discipline to use Prep. Over this 20 sec we can assume that you have a natural regen of about 10 based on gear. Prep generates 125 hatred. Overall you generate 325 hatred and only spend 260 of it. This is why I say Prep is over kill. Trying to get even more regen for either resource at this point is pointless. You have no way to burn off the extra hatred unless you are in a special case where spamming something like Strafe or Cluster Arrow is a good idea.
If for example you want to get more damage out you can switch out Prep for the Companion(Bat) you will generate less hatred but also have more damage output on top of freeing up discipline to use on other things like Vault of Sentry.
That's all well and good if all you want to do is spam an attack skill, but what if you want to do other things like Vault around a lot, drops lots of Caltrops, go invis all the time with Smoke Screen, or try to keep Shadow Power up as much as possible for 20% leach and 65% damage reduction? Even if you can get enough passive hatred regen and a slow enough attack speed to not need Prep, you are still limited by your natural Discipline regen. Multishot lets you get so much more Discipline than you could get any other way. Perhaps this is a different build--max Discipline defensive/utility that just happens to not require hatred generators, instead of a no hatred generator build that just happens to generate a ton of Discipline for defensive/utility use--but either way, it seems like a great way to build the DH.
Well yes that is the point but the original argument was that improving discipline would give more dps. Getting more disc for other things like Vault/SP and other survival options is a good idea and I see where you are going with this. Just remember the original question was if boosting your disc regen would effect your dps.
Rollback Post to RevisionRollBack
(╯°□°)╯︵ ┻━┻ h-o-p-e-s.net ┬─┬ノ( º _ ºノ) Sorry had to fix that table.
Looked at the post Exitium. The numbers check out but what I would like to point out is that no weapon is slower than 1.1 APS and this is with no +mod for APS. Remember that you will spend more as speeds go up but, aside from multi, only gain the same amount of disc and thus a fixed amount of hatred over that same time.
What I'm going to do when I get home from class is to compile a list of attack speed ranges that can be used with different hatred generators. This will give a better idea of how fast you can get with different options to help with efficiency issues people may have when building this type of build.
Rollback Post to RevisionRollBack
(╯°□°)╯︵ ┻━┻ h-o-p-e-s.net ┬─┬ノ( º _ ºノ) Sorry had to fix that table.
Yeah I was going to factor attack speeds in eventually. Since I calculated everything based on 1 APS then all you have to do is multiply the figures I gave by the APS value. i.e. EA spam @ 1APS = 10H/s, but @ 1.4APS it'd be 14H/s.
Discipline Spender Estimates
Those "Durations" don't make sense to me. Caltrops, Vault, and Prep don't have a duration or a cooldown (other than the spell cast time). Mark lasts for 30 seconds, though you can cast it more frequently on multiple targets. Smoke Screen lasts for 2 second (3 with the rune) and Shadow Power lasts for 3 seconds.
It seems like you are actually using the duration column to estimate average time between casts? I think it would be more useful to calculate the D/s required for 100% uptime on SS and SP (good to know for MS builds) and then maybe calculate average uptime at one or two common D/s values for all other builds.
Hatred Spenders
Hatred Spenders with cooldowns aren't exactly additive with other hatred usage rates. You can't fire off two attacks at the same time, so if you use RoV on cooldown in an EA build you are really only spending an extra 30H every 30 seconds. So, RoV costs a max of 1H/s and FoK costs a max of 1H/s as well.
More importantly, though, your estimate of 1 APS doesn't cover any real world situation at all. I threw together a spreadsheet to compare different weapon setups, but it also illustrates the ranges for attack speeds with these setups (without a Frenzy Shrine): https://docs.google....V2Z5aVhWZFczVWc
Here I'm looking at only attack speed and +damage modifiers from gear (+% damage from dex and skills is on top of this, but it effects all setups equally). In all cases I've assumed the top level 60 items with average base weapon damage and average quiver automod damage.
The first block of data is base values with no additional stats from gear. The second block is with max stats from gear. Next I show which gear pieces with what mods are used. Finally I threw in some tables that isolate each stat to see if the relative positions of the item setups change by focusing on one stat and no others. In nearly all case, the setups from best to worst are:
The only case where this doesn't hold is the final one with max +damage but no attack speed improvements of any kind. In this case 2h bow takes the lead with dual wield right behind, but all setups besides the shield are very close. The most interesting result to me was the poor performance of the dual wield combo. It turns out that the 15% IAS bonus stacks additively with other IAS from gear rather than multiplicatively and that IAS on a weapon effect only that weapon. A quiver can have up to 17% IAS, beating out the dual wield bonus. The one thing that keeps dual wield from falling hopelessly behind is that fixed +attack speed on a weapon applies to both weapons when dual wielding, giving that setup the possibility of an extra +0.25 attacks per second affix.
Back to the weapon speed issues, here is the possible range of speeds for each setup (numbers in parens are max speed with a Frenzy Shrine):
The slowest possible attack speed is a 2h xbow with no speed modifiers at 1.1 APS. If you want to use a 2h bow or 1h xbow then you can't get any slower than 1.4 and 1.6 APS respectively. With full attack speed gear, a dual wield setup can get up to a whopping 6.14 APS with a Frenzy Shrine and even a 2h xbow with that gear level has 3.84 APS without a shrine. Even with a 10H/attack skill, that's going to be 38.4H/s required which isn't quite sustainable even with max hatred regen gear, companion, passives, and Prep, and a Templar.
Now, obviously we shouldn't expect to hit the absolute max attack speeds, and a non-MS build can actively avoid those attack speed boosts (at the potential cost of overall dps), but it doesn't seem at all unlikely to hit 2.0+ APS. As gear improves, a non-MS build is going to have to focus more and more on hatred generation to support itself. You do, however, always have the option of switching to MS as your primary attack, which can support itself even at a whopping 6.14 APS * 15H = 92H/s (you'd have to mash that Prep button every 1-2 seconds, but with over 30D/s from MS if you're hitting 5 targets at this speed, you can certainly afford that).
Yeah I got kind of tired of typing so I didn't write it all out fully.
Basically, the Duration stuff either looks at the actually duration of an ability, assumes you will use it back-to-back, then estimates how many times you might you it in one "Duration". For abilities without a cooldown, I estimate how often they're used, i.e. my estimate of Vault usage was once per 10 seconds. You make a good point about SS and SP, but can you attack whilst SS is up? In the table I estimate you might use SS or SP once per ten seconds. The formatting was lost on those table, annoyingly. Also it doesn't seem the forum understand the greater than symbol...
The hatred spender costs increase directly dependantly on the H/s that I estimate. The post before yours shows what I mean. The post I did with the Math was a basis for me to continue to "prove" what will and won't work. I totally agree with all your points on IAS, took a quick look at the spreadsheet too. It was becoming obvious as I wrote aforementioned post that a 2h Xbow would probably be the best option for most builds - with quiver of course. The effect is actually twofold as well, since using a 2h weapon bias' you to selecting IAS gear for most DPS, it actually increases the chance for number of Nightstalker procs (see equation) AND the effect of Archery increases your Critical Strike damage. Admittedly though DW'ing is probably not far behind, so it seems things are kind of normalised to a point.
@Chippy: Having a blah kinda night so it was mostly TL;DR however I did skim it and pick out something I kinda want to say to that. You did a good job pulling all of that together but I am kinda concerned by the conclusion of the sheer absurd attack speeds. 4+ APS? I have yet to see anything in any video go this fast ever. Not for any class at any level range. I have a distinct feeling the attack speed stat is NOT additive but something else.
Directly from the numbers in the beta. 7% and 6% on 1h xbows with 15% duel wield. In total that should be an increase on the standard APS by .448. Basically 1.6+.448 getting 2.05. THIS ASSUMES IT IS ADDITIVE. When i actually equip the gear it gives me only 1.97. There is either some amount of diminishing returns or nature (or some other relation we don't have an equation for just yet) that limits attack speed. I highly doubt that we will break the 3 barrier very easily even with duel wielding.
I tired saying before that we need to balance APS with our hatred generators. We know there are limits on how much disc and hatred you gen and use the real issue is the exact build you are using and the gear you have available to you.
I'm also surprised at your damage estimates. They look way too unbalanced (I mean come on relative % diff for them ranges all the way down to 43%? ). I have a feeling you have something way off in your calculations if these relations are so far away from each other. If I was a game designer, and to be honest I am in college for it, I would take one look at those numbers and go back to the dev team and have them rework the numbers to balance it out. There really should be no reason to gimp players so baddly for using 2 1h xbows. Especially when that is the main weapon the design of all of the artwork and themeing is based around.
Rollback Post to RevisionRollBack
(╯°□°)╯︵ ┻━┻ h-o-p-e-s.net ┬─┬ノ( º _ ºノ) Sorry had to fix that table.
@Chippy: Having a blah kinda night so it was mostly TL;DR however I did skim it and pick out something I kinda want to say to that. You did a good job pulling all of that together but I am kinda concerned by the conclusion of the sheer absurd attack speeds. 4+ APS? I have yet to see anything in any video go this fast ever. Not for any class at any level range. I have a distinct feeling the attack speed stat is NOT additive but something else.
Directly from the numbers in the beta. 7% and 6% on 1h xbows with 15% duel wield. In total that should be an increase on the standard APS by .448. Basically 1.6+.448 getting 2.05. THIS ASSUMES IT IS ADDITIVE. When i actually equip the gear it gives me only 1.97. There is either some amount of diminishing returns or nature (or some other relation we don't have an equation for just yet) that limits attack speed. I highly doubt that we will break the 3 barrier very easily even with duel wielding.
Basically, though, IAS on a weapon only counts for that one weapon. In your case we have 1.6 * 1.07 * 1.15 = 1.9688 = 1.97. If you put the 6% bow in your main hand you should see 1.6 * 1.06 * 1.15 = 1.9504 = 1.95.
In the beta you're looking at max level 13 gear, which is a far cry from the stats you can get at level 60. I've seen all of the stats I used for that spreadsheet in the beta (at lower values) except for the flat +attack speed on magic items (which I've only seen on a white item as far as I can remember) and have tested the attack speed formula I used in-game and have yet to find an example which doesn't match the prediction.
If you think about it, those attack speeds aren't wildly unreasonable compared to Diablo 2. A 4-frame zeal was 6.25 APS and a 2-frame strafe was 12.5 APS. Even a 7-frame MS was 3.57 APS, so these number's don't seem wildly unreasonable compared to what was possible in D2 (and it was a lot easier to hit the max breakpoints in D2 than it will be to put together a set of max attack speed gear in D3).
I'm also surprised at your damage estimates. They look way too unbalanced (I mean come on relative % diff for them ranges all the way down to 43%? ). I have a feeling you have something way off in your calculations if these relations are so far away from each other. If I was a game designer, and to be honest I am in college for it, I would take one look at those numbers and go back to the dev team and have them rework the numbers to balance it out. There really should be no reason to gimp players so baddly for using 2 1h xbows. Especially when that is the main weapon the design of all of the artwork and themeing is based around.
First of all, the 1h xbow + shield definitely should be below the other dps numbers.
Second, the bottom 4 test cases were just thrown in there to see if the relative ordering would change if you prioritized different stats--the relative dps values for those cases are basically worthless. Stacking attack speed in its different forms is more beneficial to the slower base attack weapons than the faster ones, so its no surprise that the 1h xbow setups do poorly under these conditions. On the flip side, stacking +damage is more beneficial to the faster base attack speeds and we can see from the final test case that all weapon setups are pretty much dead even in that case.
Looking at the second set of values on top (the max possible dps setup). The dual wield setup is only 13% behind the 2h xbow setup, but has a faster attack speed if your build benefits from any sort of procs and has an extra slot that can roll the Hatred regen and max Discipline affixes. Sure it could be closer, but it doesn't seem way out of line.
I went ahead and added another test at the very bottom to show what the numbers would look like if Quivers were a true off-hand item (if they couldn't be used with a 2h weapon). In this case, the 1h xbow setups slightly edge out the 2h setups. Its very close in this comparison, but the 2h setups will fall further behind when you account for the fact that they will have a few less affixes on the one item in the weapon slots compared to two items in the 1h xbow setups. Something like this is what other classes are likely to see with the different item setups being pretty darn close. The Demon Hunter has this weird mechanic of being able to have double stats with 2h base damage at the same time. It's probably not as balanced as it could be, but even a 13% difference is likely to be the sort of on-paper thing that is virtually undetectable in the actual game (player proficiency and skill selection will account for a lot bigger difference than this).
Another thing to consider is that gear differences dwarf these variations by orders of magnitude. Going from no stats to max stats is nearly a factor of 10 difference. Once you include Dexterity thats probably nearly another factor of 20 plus possible crit, etc. All told, gear affixes will probably increase total dps by as much as 200 times what's possible with level 60 white gear. Because of this, a good pair of 1h xbows will almost certainly beat out a mediocre 2h xbow + quiver (and visa versa).
@Chippy you really should take this info to a new thread. Not saying I don't want you here it is just a wealth of information being wasted in an off topic fashion. I know it all stemmed from a question about the builds but we keep getting so far off topic and I would like to at least try to stay on topic in this thread. I implore you to take this to a new thread and show the community the work you put into this! The math is honestly amazing and the datamineing you have done far surpasses my own, it really deserves its own thread.
You make a good point at the end of the argument. We are only very narrowly factoring in two types of stats when in all reality we will see 3+ affixes per piece of gear and these numbers will most likely balance out. I did not really take into account that the sample was a very narrow look into how only attack speed effected things. Though with this point it I would guess that my reasoning of saying changing your attack speed to fit the build will not change dps much when factoring in the additional stats. E.G slowing down your APS to allow for the skillset you have chosen to work or speeding it up to be more efficient with your hatred. Remember those are only small changes in attack speed and as long as you are under about 27 hatred spent per second you can make these builds work to some level of efficiency.
Edit: I only say 27 because this is the highest I have been able to figure with napkin math hatred gen per sec can get using this type of build.
Rollback Post to RevisionRollBack
(╯°□°)╯︵ ┻━┻ h-o-p-e-s.net ┬─┬ノ( º _ ºノ) Sorry had to fix that table.
Edit: I only say 27 because this is the highest I have been able to figure with napkin math hatred gen per sec can get using this type of build.
I think it's possible to get more H/s than that. It revolves around Multishotting and Prep (Pun) - remembering that Prep has no CD in that form, it means you can theoretically generate "infinite" hatred - though it's not really infinite because it takes time to cast MS.
In fact, I'll put some numbers into some of my equations and see what happens... Let's say we have an AS of 3/sec, so we want to be aiming at about 45H/s to be totally super-spam sustainable.
We'll first need to estimate a max D regen, based on the MS equation I made: D/s = 5N/T + B. We know that we can "cast" 3 MS's per second here, if we cast MS pretty much non-stop we'll generate 16D/s, which would mean we could use Prep every second if we had Perfection also... Gasp!
H/s = V(Dr/Dc) + B
V here is 150, if we assume Vengeance, but lets also assume Perfection, Companion.and 6H/s from gear. Hence, Dc is 16, B is 13.5.
Sub in our target of 45; 45 = 150(Dr/16) + 13.5.
Rearrage and solve gives Dr = 3.36 D/s needed to sustain. i.e. we need to cast Preparation once every 4.8secs. That being said, this wouldn't leave us with any Discipline for anything else, such as Caltrops or Vault. If we had another 2D/s to account for Caltrops and Vault then we want a regen of 5.4D/s - still casting Prep once every 4.8 secs. This is getting into the realms of super button mashing though - I won't be aiming towards anything as fast paced as this - unless I need to to be competitive.
To get 5.4D/s we would need to cast N = [(5.4 - B)T]/5, where B is 1 and T is 4.8. i.e. N = 4.2 Multishots per cycle, This would take 1.4 seconds of our 4.8 seconds cycle. Leaving 3.4 secs, i.e. 10 "cast" slots - 10x15 = 150. This checks out since we're generating 45H/s (think of it as Prep covering all the damage and passives as covering MS). - That was just to check I hadn't done a bad.
Please scrutinise.
Basically it is possible, you could even couple Nightstalker and Multishot within a crit build. This is also very useful. A skill list for all classes and abilities is here. There are affixes for RF (crit) and EA (damage).
Edit: Just noticed that when I factored Vault/Caltrops in, we wouldn't actually be casting Prep any more often - the numbers have been changed accordingly.
Possible yes but I was only factoring things that were guaranteed we only know the rates of some hatred/disc returns on attacks. Night Stalker may have good potential but since we don't know the proc rate and the average crit level players will have it is impossible at this point to tell if it is a good passive for these builds or not.
Also my issue with multi is that it is very short range. 15? (b.net is down so I can't find the actual value) yards is short. I've had this exact convo in the monk forums about another skill that was 15 yard range that people were talking about as if it could hit anything on screen. Problem is none of them had access to the beta and this is far from the case. 15 yards is surprisingly short. For a DH who has absolutely no good means of damage mitigation other than avoidance and slows to get in close to use this sort of build will be highly deadly.
Where a wizard can freeze, armor and has damage absorption and the WD has several ways to reduce damage and heal we have two passives. One only works after 5 sec of damage and the other is based off of calt/spike trap. Problem is one is a yellow skill and caltrops drop at your feet not at the cursor location. By the time they reach this spot you would have already had to move away from the position. DH, unlike pretty much every other class, are very paper thin when it comes to defense. We are more of a glass cannon than a Wizard who is trying for the same sort of close in setup.
This is exactly why I say min/maxing is bad after a certain point. Yes you can get the numbers to say something crazy but in all reality what you have to do to achieve these numbers is very hard/dangerous. Even if you can justify to use this build you will have to spend quite a lot of time running away from targets to stay alive and thus lowering your damage.
Also I was trying to see how high one could get hatred regen to see if Cluster was spammable. There is no way I can see to do that it just costs too much. The next highest to spam is Impale and that is quite a good option in some of the higher end hatred generation builds (also costing half of what Cluster does has benefits)
Rollback Post to RevisionRollBack
(╯°□°)╯︵ ┻━┻ h-o-p-e-s.net ┬─┬ノ( º _ ºノ) Sorry had to fix that table.
A good point about the range. Still, this is assuming 3APS, which if using a 2h Xbow is pretty hard to get from 1.1APS. (272%IAS). Like I said in the post, I'm not aiming to get that much IAS for any build, I'm just showing that the setup still works even at crazy levels.
@Chippy you really should take this info to a new thread. Not saying I don't want you here it is just a wealth of information being wasted in an off topic fashion. I know it all stemmed from a question about the builds but we keep getting so far off topic and I would like to at least try to stay on topic in this thread. I implore you to take this to a new thread and show the community the work you put into this! The math is honestly amazing and the datamineing you have done far surpasses my own, it really deserves its own thread.
Yeah, sorry...I was thinking I was starting to go off-topic, but didn't really feel like starting a new thread.
Possible yes but I was only factoring things that were guaranteed we only know the rates of some hatred/disc returns on attacks. Night Stalker may have good potential but since we don't know the proc rate and the average crit level players will have it is impossible at this point to tell if it is a good passive for these builds or not.
As of now, max crit% from gear is 54% plus 5% base plus 10% from Archery with hand xbows plus 10% from a Scoundrel follower. That's enough to get you to what I've heard is the hard-cap of 75%. Again, standard caveats that you actually have to find the gear and that we don't know what the proc rate is yet. With 75% crit you'd be getting anywhere from 0-1.5D/hit depending on proc rate. So, its possible that you could have the MS Discipline generation benefits with any multi-target skill this way.
Also my issue with multi is that it is very short range. 15? (b.net is down so I can't find the actual value) yards is short. I've had this exact convo in the monk forums about another skill that was 15 yard range that people were talking about as if it could hit anything on screen. Problem is none of them had access to the beta and this is far from the case. 15 yards is surprisingly short. For a DH who has absolutely no good means of damage mitigation other than avoidance and slows to get in close to use this sort of build will be highly deadly.
Not sure where you got this 15 yard number from, but I can assure you its wrong. You can see a video of MS in action on the official skill page: http://us.battle.net/d3/en/class/demon-hunter/active/multishot. The shot seems to be hitting nearly to the edge of the screen which is at least 35 yards (roughly Vault distance). Perhaps 15 yards has something to do with how wide the shot hits?
As for mitigation, MS should give enough Discipline regen in most cases to allow very high uptime on Shadow Power w/ Gloom. That's 65% damage reduction with 20% life steal, which seems like it should be plenty tanky to let you go toe-to-toe when you want to.
@Chippy Well my point about night stalker was that it was a chance on crit. It can be something like internal CD of x amount of time until it can happen again, or something closer to 10% chance to proc on crit. It does look powerful but like I said it is something that is hard for us to judge. Yes it can be as good as you said but the problem is that it can also be quite bad. The random part of the variable is the reason why I did not add it into the calculations.
Lots of those procs are handled in wow by ICD so that probably is not all too far from what they may have implemented here.
Also nice find on the vid of multi. I never noticed that it had such a good range. The last time we saw the spell it was highly range limited. It really makes me wonder why they have the rune that generates discipline on it in the first place. Even hitting 1 target is good enough to consider using multi over other attacks. If they never change multi and leave that rune, which they may put some type of limit on it sometime in the future, it looks highly overpowered when taking all the discipline options it opens up for any type of but not only these. I am kinda shocked that they would bump up the range AND remove the 6 max target off the rune. It may be the most useful attack in the whole skill sheet for us.
Edit: don't say sorry. The work you did is awesome.
Rollback Post to RevisionRollBack
(╯°□°)╯︵ ┻━┻ h-o-p-e-s.net ┬─┬ノ( º _ ºノ) Sorry had to fix that table.
There are so many ways to have great discipline and heatred regen with DH and all of them will depend on DH level and gear. I think it will be a class which we will have to change very often when we progress. I love it already, though I don't even have beta yet. It's probably gonna be my first character and then some melee.
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
We're trying to lessen Discipline consumption whilst being able to maintain the "spam offensive" idea, this leaves us more breathing room when we want to use other Discipline abilities - without being tied down to X number of Multishots per cycle. Though admittedly it's the simplest solution. My "guide" post was to look at the possibilities we had... Though unfortunately it needs revising, which I think I'll do now.
Edit: And as azure says, we want to optimise how much hatred we actually generate, it's kind of min/maxing within the build.
P.S. I'm not doing this so I can pick which build style is the best dps, I want to play the most fun one... The more options there are the better the chance I'll come up with something great!
H Generation Options: ¦ D Generation Options
Abilities are assumed to be used exactly on cooldown.
No Passives etc are included in calculations unless otherwise specified.
D regeneration is assumed to be base, i.e. One.
This formula is used for H Generators marked with an * :
H/s = V(Dr/Dc) + B where V is the amount of Hatred the skill grants , Dr is the Discipline Regen, Dc is the Discipline cost of the skill. B is the Sum of True Passive Regens (i.e "Base" regen + Gear).
1) Prep ; 10.75H/s with No Veng. 12H/s with Veng. 13.9H/s with Veng & Perf.
2) Gear (Pure) - DualWield + Cloak + Base = 14.5H/s. 2H/Quiver + Cloak = 11.2 H/s
3) Smoke Screen ; 7H/s no Perf. 7.8H/s with Perf.
4) Shadow Power ; 6.75H/s no Perf. 7.3 with Perf.
5) Companion ; 3H/s + 4.5H/s = 7.5H/s.
Rank with no Passives: DW Pure Gear > Prep > 2H Pure Gear > Comp > SS > SP.
It is important to note that to calculate the total H generation if multiple skills are used then you must account for the fact that base regen is included in each one.
"Passive" Discipline Regeneration of each Skill
These equations/values are simple estimate as to what D regeneration we will get depending on certain gear factors and assumptions. The values will be used later on after evaluating Discipline consumption for a particular build.
1) Multishot - If we assume 5 targets are hit per use, then this gives us 5D. Assuming 1 Attack per Second, that's 5D/s. BUT, we aren't going to be spamming MS constantly - it's better to evaulate the average D/s generated over a certain time. the equation for this is:
D/s = 5N/T + B where is the Number of MS uses, T is a given time and B is the base D regen of 1/s.
For example, if we assume 3 MS uses per 15 seconds then that gives us a total regen of about 2D/s.
2) Nightstalker - this is very difficult to accurately quantify because it depends on several variables. I worked out an equation that provides a rough estimate of D/s, such that if we gain more information it will become useful. As is, it is a vague estimate based on educated guesswork. It is:
D/s = 2CP(APSbar) C is Crit chance and P is Proc Chance, expressed as decimals ideally. APSbar is the Average Attacks per Second.
So, if we had 35% Crit, the Proc chance of NS is 50% and we average 5APS then D/s is 2(0.35)(0.5)(5) = 1.75.
3) Prep (unruned) - 30D/45s = 0.66D/s
4) Prep (Focused Mind) - 45D/45s = 1D/s (I believe Cooldown remains), providing you don't cap D whilst Prep is active.
Naturally you can't use Prep for D regen if you're using it for H regen, but if you do use it for D regen you should rune it unless you prefer the instant effect of the unruned version.
Discipline Spenders - Estimations!
I've estimated how often a skill might get used if it has no duration, if it has a duration I have assumed it's used back-to-back. Sentry & Companion have been excluded, Sentry because I'm not sure if you can have multiple ones down, Companion because it's one use only.
Skill - Cost - Cost /w Perf - Duration - Usage/per Duration
Skill - Estimate of D/s - Estimate of D/s with Perf
Using these figures I've estimated rough discipline consumptions for certain build styles:
Hatred Spenders - Estimations
Both Cluster Arrow and Impale have been excluded from my estimations because I believe they are both special cases that require analysis within a particular build.
Assuming 1 APS and spamming ability, or use on cooldown:
Now we have rough estimates for a particular build, or what sort of generation we need - we also have estimates of how much regeneration a given ability provides.
More to come later on/Tommorrow.
That's all well and good if all you want to do is spam an attack skill, but what if you want to do other things like Vault around a lot, drops lots of Caltrops, go invis all the time with Smoke Screen, or try to keep Shadow Power up as much as possible for 20% leach and 65% damage reduction? Even if you can get enough passive hatred regen and a slow enough attack speed to not need Prep, you are still limited by your natural Discipline regen. Multishot lets you get so much more Discipline than you could get any other way. Perhaps this is a different build--max Discipline defensive/utility that just happens to not require hatred generators, instead of a no hatred generator build that just happens to generate a ton of Discipline for defensive/utility use--but either way, it seems like a great way to build the DH.
Well yes that is the point but the original argument was that improving discipline would give more dps. Getting more disc for other things like Vault/SP and other survival options is a good idea and I see where you are going with this. Just remember the original question was if boosting your disc regen would effect your dps.
h-o-p-e-s.net
┬─┬ノ( º _ ºノ)
Sorry had to fix that table.
What I'm going to do when I get home from class is to compile a list of attack speed ranges that can be used with different hatred generators. This will give a better idea of how fast you can get with different options to help with efficiency issues people may have when building this type of build.
h-o-p-e-s.net
┬─┬ノ( º _ ºノ)
Sorry had to fix that table.
Those "Durations" don't make sense to me. Caltrops, Vault, and Prep don't have a duration or a cooldown (other than the spell cast time). Mark lasts for 30 seconds, though you can cast it more frequently on multiple targets. Smoke Screen lasts for 2 second (3 with the rune) and Shadow Power lasts for 3 seconds.
It seems like you are actually using the duration column to estimate average time between casts? I think it would be more useful to calculate the D/s required for 100% uptime on SS and SP (good to know for MS builds) and then maybe calculate average uptime at one or two common D/s values for all other builds.
Hatred Spenders
Hatred Spenders with cooldowns aren't exactly additive with other hatred usage rates. You can't fire off two attacks at the same time, so if you use RoV on cooldown in an EA build you are really only spending an extra 30H every 30 seconds. So, RoV costs a max of 1H/s and FoK costs a max of 1H/s as well.
More importantly, though, your estimate of 1 APS doesn't cover any real world situation at all. I threw together a spreadsheet to compare different weapon setups, but it also illustrates the ranges for attack speeds with these setups (without a Frenzy Shrine): https://docs.google....V2Z5aVhWZFczVWc
Here I'm looking at only attack speed and +damage modifiers from gear (+% damage from dex and skills is on top of this, but it effects all setups equally). In all cases I've assumed the top level 60 items with average base weapon damage and average quiver automod damage.
The first block of data is base values with no additional stats from gear. The second block is with max stats from gear. Next I show which gear pieces with what mods are used. Finally I threw in some tables that isolate each stat to see if the relative positions of the item setups change by focusing on one stat and no others. In nearly all case, the setups from best to worst are:
2h xbow + quiver
2h bow + quiver
1h xbow + quiver
1h xbow + 1h xbow
1h xbow + shield
The only case where this doesn't hold is the final one with max +damage but no attack speed improvements of any kind. In this case 2h bow takes the lead with dual wield right behind, but all setups besides the shield are very close. The most interesting result to me was the poor performance of the dual wield combo. It turns out that the 15% IAS bonus stacks additively with other IAS from gear rather than multiplicatively and that IAS on a weapon effect only that weapon. A quiver can have up to 17% IAS, beating out the dual wield bonus. The one thing that keeps dual wield from falling hopelessly behind is that fixed +attack speed on a weapon applies to both weapons when dual wielding, giving that setup the possibility of an extra +0.25 attacks per second affix.
Back to the weapon speed issues, here is the possible range of speeds for each setup (numbers in parens are max speed with a Frenzy Shrine):
2h xbow + quiver 1.10 - 3.84 (4.36)
2h bow + quiver 1.40 - 4.53 (5.14)
1h xbow + quiver 1.60 - 5.99 (5.67)
1h xbow + 1h xbow 1.84 - 5.40 (6.14)
1h xbow + shield 1.60 - 4.54 (5.21)
The slowest possible attack speed is a 2h xbow with no speed modifiers at 1.1 APS. If you want to use a 2h bow or 1h xbow then you can't get any slower than 1.4 and 1.6 APS respectively. With full attack speed gear, a dual wield setup can get up to a whopping 6.14 APS with a Frenzy Shrine and even a 2h xbow with that gear level has 3.84 APS without a shrine. Even with a 10H/attack skill, that's going to be 38.4H/s required which isn't quite sustainable even with max hatred regen gear, companion, passives, and Prep, and a Templar.
Now, obviously we shouldn't expect to hit the absolute max attack speeds, and a non-MS build can actively avoid those attack speed boosts (at the potential cost of overall dps), but it doesn't seem at all unlikely to hit 2.0+ APS. As gear improves, a non-MS build is going to have to focus more and more on hatred generation to support itself. You do, however, always have the option of switching to MS as your primary attack, which can support itself even at a whopping 6.14 APS * 15H = 92H/s (you'd have to mash that Prep button every 1-2 seconds, but with over 30D/s from MS if you're hitting 5 targets at this speed, you can certainly afford that).
Basically, the Duration stuff either looks at the actually duration of an ability, assumes you will use it back-to-back, then estimates how many times you might you it in one "Duration". For abilities without a cooldown, I estimate how often they're used, i.e. my estimate of Vault usage was once per 10 seconds. You make a good point about SS and SP, but can you attack whilst SS is up? In the table I estimate you might use SS or SP once per ten seconds. The formatting was lost on those table, annoyingly. Also it doesn't seem the forum understand the greater than symbol...
The hatred spender costs increase directly dependantly on the H/s that I estimate. The post before yours shows what I mean. The post I did with the Math was a basis for me to continue to "prove" what will and won't work. I totally agree with all your points on IAS, took a quick look at the spreadsheet too. It was becoming obvious as I wrote aforementioned post that a 2h Xbow would probably be the best option for most builds - with quiver of course. The effect is actually twofold as well, since using a 2h weapon bias' you to selecting IAS gear for most DPS, it actually increases the chance for number of Nightstalker procs (see equation) AND the effect of Archery increases your Critical Strike damage. Admittedly though DW'ing is probably not far behind, so it seems things are kind of normalised to a point.
Directly from the numbers in the beta. 7% and 6% on 1h xbows with 15% duel wield. In total that should be an increase on the standard APS by .448. Basically 1.6+.448 getting 2.05. THIS ASSUMES IT IS ADDITIVE. When i actually equip the gear it gives me only 1.97. There is either some amount of diminishing returns or nature (or some other relation we don't have an equation for just yet) that limits attack speed. I highly doubt that we will break the 3 barrier very easily even with duel wielding.
I tired saying before that we need to balance APS with our hatred generators. We know there are limits on how much disc and hatred you gen and use the real issue is the exact build you are using and the gear you have available to you.
I'm also surprised at your damage estimates. They look way too unbalanced (I mean come on relative % diff for them ranges all the way down to 43%? ). I have a feeling you have something way off in your calculations if these relations are so far away from each other. If I was a game designer, and to be honest I am in college for it, I would take one look at those numbers and go back to the dev team and have them rework the numbers to balance it out. There really should be no reason to gimp players so baddly for using 2 1h xbows. Especially when that is the main weapon the design of all of the artwork and themeing is based around.
h-o-p-e-s.net
┬─┬ノ( º _ ºノ)
Sorry had to fix that table.
1.97 is exactly what I would expect from that setup. I posted a separate thread with the full attack speed formula here: http://www.diablofans.com/topic/36153-attack-speed-formula/
Basically, though, IAS on a weapon only counts for that one weapon. In your case we have 1.6 * 1.07 * 1.15 = 1.9688 = 1.97. If you put the 6% bow in your main hand you should see 1.6 * 1.06 * 1.15 = 1.9504 = 1.95.
In the beta you're looking at max level 13 gear, which is a far cry from the stats you can get at level 60. I've seen all of the stats I used for that spreadsheet in the beta (at lower values) except for the flat +attack speed on magic items (which I've only seen on a white item as far as I can remember) and have tested the attack speed formula I used in-game and have yet to find an example which doesn't match the prediction.
If you think about it, those attack speeds aren't wildly unreasonable compared to Diablo 2. A 4-frame zeal was 6.25 APS and a 2-frame strafe was 12.5 APS. Even a 7-frame MS was 3.57 APS, so these number's don't seem wildly unreasonable compared to what was possible in D2 (and it was a lot easier to hit the max breakpoints in D2 than it will be to put together a set of max attack speed gear in D3).
First of all, the 1h xbow + shield definitely should be below the other dps numbers.
Second, the bottom 4 test cases were just thrown in there to see if the relative ordering would change if you prioritized different stats--the relative dps values for those cases are basically worthless. Stacking attack speed in its different forms is more beneficial to the slower base attack weapons than the faster ones, so its no surprise that the 1h xbow setups do poorly under these conditions. On the flip side, stacking +damage is more beneficial to the faster base attack speeds and we can see from the final test case that all weapon setups are pretty much dead even in that case.
Looking at the second set of values on top (the max possible dps setup). The dual wield setup is only 13% behind the 2h xbow setup, but has a faster attack speed if your build benefits from any sort of procs and has an extra slot that can roll the Hatred regen and max Discipline affixes. Sure it could be closer, but it doesn't seem way out of line.
I went ahead and added another test at the very bottom to show what the numbers would look like if Quivers were a true off-hand item (if they couldn't be used with a 2h weapon). In this case, the 1h xbow setups slightly edge out the 2h setups. Its very close in this comparison, but the 2h setups will fall further behind when you account for the fact that they will have a few less affixes on the one item in the weapon slots compared to two items in the 1h xbow setups. Something like this is what other classes are likely to see with the different item setups being pretty darn close. The Demon Hunter has this weird mechanic of being able to have double stats with 2h base damage at the same time. It's probably not as balanced as it could be, but even a 13% difference is likely to be the sort of on-paper thing that is virtually undetectable in the actual game (player proficiency and skill selection will account for a lot bigger difference than this).
Another thing to consider is that gear differences dwarf these variations by orders of magnitude. Going from no stats to max stats is nearly a factor of 10 difference. Once you include Dexterity thats probably nearly another factor of 20 plus possible crit, etc. All told, gear affixes will probably increase total dps by as much as 200 times what's possible with level 60 white gear. Because of this, a good pair of 1h xbows will almost certainly beat out a mediocre 2h xbow + quiver (and visa versa).
You make a good point at the end of the argument. We are only very narrowly factoring in two types of stats when in all reality we will see 3+ affixes per piece of gear and these numbers will most likely balance out. I did not really take into account that the sample was a very narrow look into how only attack speed effected things. Though with this point it I would guess that my reasoning of saying changing your attack speed to fit the build will not change dps much when factoring in the additional stats. E.G slowing down your APS to allow for the skillset you have chosen to work or speeding it up to be more efficient with your hatred. Remember those are only small changes in attack speed and as long as you are under about 27 hatred spent per second you can make these builds work to some level of efficiency.
Edit: I only say 27 because this is the highest I have been able to figure with napkin math hatred gen per sec can get using this type of build.
h-o-p-e-s.net
┬─┬ノ( º _ ºノ)
Sorry had to fix that table.
I think it's possible to get more H/s than that. It revolves around Multishotting and Prep (Pun) - remembering that Prep has no CD in that form, it means you can theoretically generate "infinite" hatred - though it's not really infinite because it takes time to cast MS.
In fact, I'll put some numbers into some of my equations and see what happens... Let's say we have an AS of 3/sec, so we want to be aiming at about 45H/s to be totally super-spam sustainable.
We'll first need to estimate a max D regen, based on the MS equation I made: D/s = 5N/T + B. We know that we can "cast" 3 MS's per second here, if we cast MS pretty much non-stop we'll generate 16D/s, which would mean we could use Prep every second if we had Perfection also... Gasp!
H/s = V(Dr/Dc) + B
V here is 150, if we assume Vengeance, but lets also assume Perfection, Companion.and 6H/s from gear. Hence, Dc is 16, B is 13.5.
Sub in our target of 45; 45 = 150(Dr/16) + 13.5.
Rearrage and solve gives Dr = 3.36 D/s needed to sustain. i.e. we need to cast Preparation once every 4.8secs. That being said, this wouldn't leave us with any Discipline for anything else, such as Caltrops or Vault. If we had another 2D/s to account for Caltrops and Vault then we want a regen of 5.4D/s - still casting Prep once every 4.8 secs. This is getting into the realms of super button mashing though - I won't be aiming towards anything as fast paced as this - unless I need to to be competitive.
To get 5.4D/s we would need to cast N = [(5.4 - B)T]/5, where B is 1 and T is 4.8. i.e. N = 4.2 Multishots per cycle, This would take 1.4 seconds of our 4.8 seconds cycle. Leaving 3.4 secs, i.e. 10 "cast" slots - 10x15 = 150. This checks out since we're generating 45H/s (think of it as Prep covering all the damage and passives as covering MS). - That was just to check I hadn't done a bad.
Please scrutinise.
Basically it is possible, you could even couple Nightstalker and Multishot within a crit build. This is also very useful. A skill list for all classes and abilities is here. There are affixes for RF (crit) and EA (damage).
Edit: Just noticed that when I factored Vault/Caltrops in, we wouldn't actually be casting Prep any more often - the numbers have been changed accordingly.
Also my issue with multi is that it is very short range. 15? (b.net is down so I can't find the actual value) yards is short. I've had this exact convo in the monk forums about another skill that was 15 yard range that people were talking about as if it could hit anything on screen. Problem is none of them had access to the beta and this is far from the case. 15 yards is surprisingly short. For a DH who has absolutely no good means of damage mitigation other than avoidance and slows to get in close to use this sort of build will be highly deadly.
Where a wizard can freeze, armor and has damage absorption and the WD has several ways to reduce damage and heal we have two passives. One only works after 5 sec of damage and the other is based off of calt/spike trap. Problem is one is a yellow skill and caltrops drop at your feet not at the cursor location. By the time they reach this spot you would have already had to move away from the position. DH, unlike pretty much every other class, are very paper thin when it comes to defense. We are more of a glass cannon than a Wizard who is trying for the same sort of close in setup.
This is exactly why I say min/maxing is bad after a certain point. Yes you can get the numbers to say something crazy but in all reality what you have to do to achieve these numbers is very hard/dangerous. Even if you can justify to use this build you will have to spend quite a lot of time running away from targets to stay alive and thus lowering your damage.
Also I was trying to see how high one could get hatred regen to see if Cluster was spammable. There is no way I can see to do that it just costs too much. The next highest to spam is Impale and that is quite a good option in some of the higher end hatred generation builds (also costing half of what Cluster does has benefits)
h-o-p-e-s.net
┬─┬ノ( º _ ºノ)
Sorry had to fix that table.
Yeah, sorry...I was thinking I was starting to go off-topic, but didn't really feel like starting a new thread.
As of now, max crit% from gear is 54% plus 5% base plus 10% from Archery with hand xbows plus 10% from a Scoundrel follower. That's enough to get you to what I've heard is the hard-cap of 75%. Again, standard caveats that you actually have to find the gear and that we don't know what the proc rate is yet. With 75% crit you'd be getting anywhere from 0-1.5D/hit depending on proc rate. So, its possible that you could have the MS Discipline generation benefits with any multi-target skill this way.
Not sure where you got this 15 yard number from, but I can assure you its wrong. You can see a video of MS in action on the official skill page: http://us.battle.net/d3/en/class/demon-hunter/active/multishot. The shot seems to be hitting nearly to the edge of the screen which is at least 35 yards (roughly Vault distance). Perhaps 15 yards has something to do with how wide the shot hits?
As for mitigation, MS should give enough Discipline regen in most cases to allow very high uptime on Shadow Power w/ Gloom. That's 65% damage reduction with 20% life steal, which seems like it should be plenty tanky to let you go toe-to-toe when you want to.
Lots of those procs are handled in wow by ICD so that probably is not all too far from what they may have implemented here.
Also nice find on the vid of multi. I never noticed that it had such a good range. The last time we saw the spell it was highly range limited. It really makes me wonder why they have the rune that generates discipline on it in the first place. Even hitting 1 target is good enough to consider using multi over other attacks. If they never change multi and leave that rune, which they may put some type of limit on it sometime in the future, it looks highly overpowered when taking all the discipline options it opens up for any type of but not only these. I am kinda shocked that they would bump up the range AND remove the 6 max target off the rune. It may be the most useful attack in the whole skill sheet for us.
Edit: don't say sorry. The work you did is awesome.
h-o-p-e-s.net
┬─┬ノ( º _ ºノ)
Sorry had to fix that table.