Diablo® III

Mechanics And Game Information Compendium

(Sticky)

Hey apo, this is an amazing compilation of data. Well done.

Also thanks for mentioning my research in certain parts. Maybe because I´m lazy or maybe because I´d rather spend my time playing the game and being busy researching game mechanics and skills of all the classes, I haven´t been able to write this sort of compilation over in the Barb forums. Some day maybe hehe, this is the closest we have (basically just a list of my posts about certain Barb skills and game mechanics)...

http://us.battle.net/d3/en/forum/topic/8891939945

I guess I won´t have to reasearch Wizard skills with you around anymore ;) although recently I began playing the class more, trying out skills and stuff, it´s quite fun.

- If you are doing LoH tests, don't let yourself be fooled by the life return display. Life returns are displayed as a sum of all the HP you got during the last 0.5-0.857 seconds. They don't represent individual LoH ticks and the rate or value of those display ticks is not constant. It changes with life return, attack speed and sometimes randomly.
So for example if you want to test the proc coefficient of a DoT, you need to hit a a1 normal mp0 zombie that dies after one tick. Look at the life return: That's your base value for one tick. After that you can switch to a more durable monster. Add up all the life returns you get for one application and divide by the base value. Then you have the number of ticks for that DoT.

- Same goes for damage ticks in channeled spells. I have mentioned this before and I will again later, but some (most for Wizards) channeled spells and DoT effects won't display damage ticks individually but as a sum of all damage done over the last 0.5 seconds. So be careful with counting stuff on screen. It doesn't work very often.


Advanced research as a Wiz:
-> low damage white weapon, +min damage jewelry which brings minimum damage higher than maximum (if min>max the game will use min=max as the new base weapon damage so you get steady values)
-> get rid of all life steal and life regen, only use LoH if you want to test LoH
-> get hit (don´t be at max health) when testing LoH -
-> Frost Nova the monster to have a short window for testing without the risk of being killed - note that MP0-1 might be best to not get one shot as you´re not killing stuff with your <150 weapon damage weapon
-> first test should be with a 2H 1.00 aps weapon to get a basic understanding of the skill mechanics
-> put the skill you´re testing on a keyboard button, because the game doesn´t visually display mouse clicks so you wouldn´t be able to determine when the "click" occured
-> the click (skill activation) is usually displayed as a 5 frame long darkening of the skill icon, which you´ll recognize when viewing the video frame by frame
-> if you research LoH, put your mouse cursor on your health globe to record health changes
-> if you research resource regen/skill cost mechanics, the mouse cursor should be on the resource globe - note that in both cases, you may have to adjust your positioning to hit a target in the direction of either globe
-> most recording software displays ingame frame rate, if not use ctrl+r (built in game frame rate) - to have the most accurate results wait until you have a steady 60fps rate, at least for a couple of seconds - if it´s not steady the frame lengths jump around and may make it very difficult to figure out the mechanics
-> record the video and then view it frame by frame
-> if you lack patience, don´t research game mechanics

I probably forgot a few things but those tips should give you a general idea on how to make it right.

Anyway, I might be able to help you out a bit in certain aspects. I call this level 2 research because for certain skills I actually check actual monster health values each frame.

Last but not least I should tell you what exactly breakpoints apply to.

First of all, as I have said before, not every spell scales with attack speed. For example Blizzard does not. The casting animation does, but the DoT effect itself just ticks 4 times per second, regardless of your attack speed.

Other than that, there are basically two kinds of DoTs/channeled spells. Those that display all damage ticks as they happen (you can actually see the damage numbers pop for each tick individually - Arcane Torrent, Electrocute) and those that sum up all damage in 0.5 seconds intervals (Disintegrate, Ray of Frost, Archon Disintegration Wave, Energy Twister DoT).

The first group uses the above formulas for everything that is hit-related.
Your DPS is a function of damage per tick and ticks per second. So breakpoints also directly affect your damage output.
LoH procs for each tick. Every tick can crit individually and APoC procs for each critical tick. I'm not entirely sure (see "Unsolved Mysteries"), but CM and SA probably also proc for each crit tick as per the formula above.
Prime examples for this are Barbarian's Whirlwind and Run like the Wind, but also Wizard's Arcane Torrent. For those skills, it makes A LOT of sense to optimize for breakpoints, because everything is affected.

The second group works with a different Damage/DPS scaling model, which is currently unknown. For example Disintegrate displays summed up damage values every 0.5 seconds. Increasing your attack speed by even the slightest bit directly increases DPS, no matter if you are close to a breakpoint or not. CM and SA proc a lot more than the breakpoint formulas predict (see "Unsolved Mysteries").
LoH on the other hand strictly scales according to the formulas above. If you are at 2.51 APS and increase it to 2.52 APS, you will see increased DPS, but LoH stays exactly the same. If you go from 2.5 to 2.51, you will see just a minimal linear DPS increase, but a considerable bump in LoH for passing the breakpoint.
So for those spells there's still a lot of uncertainty.

In the "Skill-Specific Mechanics and Info" section below, you can find more info on Which spells use which breakpoints for what.


I believe the best way to describe skills is by separating each aspect of it:
- damage and its breakpoints, if any
- resource cost and its breakpoints, if any
- on attack, on hit and on crit procs and LoH

Let´s take for example the latest skill I´ve researched, Sleet Storm. By looking at monster health values above their heads each frame and later my health bar when testing LoH, I was able to figure out the following:

1. skill damage = aps * 6.0667% (364%/60) weapon damage each frame (aps * 364% / sec)
-> damage is applied each frame, displayed as a damage tick each 30 frames
-> aps directly increases the actual damage (value) not the frequency of how often damage is applied
-> first damage tick occurs 5-6 frames after the skill activation (skill activation is frames 0-5)

2. skill cost = 8 arcane power (tooltip/2) * (floor(30/aps))
-> resource globe responses have different scaling than damage responses and occur separately

3A) LoH = 0.1875 * (floor(30/aps))
-> each LoH tick equals a health globe response
-> the initial LoH tick occurs on the initial damage tick but then follows its own scaling
-> LoH ticks follow the same pattern as skill cost responds which means they happen simultaneously

3.B) On attack effects
-> only have a chance to proc (coefficient chance) on the initial tick

3.C) On crit effects (APoC) = 0.1875 * (floor(30/aps)) ???
->

attacks per second - proc length - frequency/sec - cost/sec
0.88236-0.90909 aps - 33 frames - 1.82 procs - 14.55 AP
0.90910-0.93750 aps - 32 frames - 1.88 procs - 15.00 AP
0.93751-0.96774 aps - 31 frames - 1.94 procs - 15.48 AP
0.96775-1.00000 aps - 30 frames - 2.00 procs - 16.00 AP
1.00001-1.03448 aps - 29 frames - 2.07 procs - 16.55 AP
1.03449-1.07142 aps - 28 frames - 2.14 procs - 17.14 AP
1.07143-1.11111 aps - 27 frames - 2.22 procs - 17.78 AP
1.11112-1.15384 aps - 26 frames - 2.31 procs - 18.46 AP
1.15385-1.20000 aps - 25 frames - 2.40 procs - 19.20 AP
1.20001-1.25000 aps - 24 frames - 2.50 procs - 20.00 AP
1.25001-1.30434 aps - 23 frames - 2.61 procs - 20.87 AP
1.30435-1.36363 aps - 22 frames - 2.73 procs - 21.82 AP
1.36364-1.42857 aps - 21 frames - 2.86 procs - 22.86 AP
1.42858-1.50000 aps - 20 frames - 3.00 procs - 24.00 AP
1.50001-1.57894 aps - 19 frames - 3.16 procs - 25.26 AP
1.57895-1.66666 aps - 18 frames - 3.33 procs - 26.67 AP
1.66667-1.76470 aps - 17 frames - 3.53 procs - 28.24 AP
1.76471-1.87500 aps - 16 frames - 3.75 procs - 30.00 AP
1.87501-2.00000 aps - 15 frames - 4.00 procs - 32.00 AP
2.00001-2.14285 aps - 14 frames - 4.29 procs - 34.29 AP
2.14286-2.30769 aps - 13 frames - 4.62 procs - 36.92 AP
2.30770-2.50000 aps - 12 frames - 5.00 procs - 40.00 AP
2.50001-2.72727 aps - 11 frames - 5.45 procs - 43.64 AP
2.72728-3.00000 aps - 10 frames - 6.00 procs - 48.00 AP
3.00001-3.33333 aps - 09 frames - 6.67 procs - 53.33 AP
3.33334-3.75000 aps - 08 frames - 7.50 procs - 60.00 AP
3.75001-4.28571 aps - 07 frames - 8.57 procs - 68.57 AP

Note that I wasn´t able to thoroughly test APoC with Sleet Storm so I´m only assuming it works similar to LoH. I´ve read that people posted getting much more APoC procs than they expected, although I didn´t see exact test results so I´ll have to somehow figure it out.

The problem here is the fact that damage has a different scaling than cost and LoH so we don´t know if the game only considers the one frame when it applies the LoH/AP cost tick to determine if APoC is triggered.

By looking at the monster health values ticking each frame I was able to see that crits occur in sets. Note that it´s almost impossible to have a steady framerate to have perfectly accurate results but so far it seems that Sleet Storm crits follow a certain pattern.

Let´s say we´re channeling Sleet Storm at exactly 2.00 aps, which translates to 15 frames per AP/LoH response. The damage is applied each frame, but if the skill crits, you supposedly get 15 frames that crit subsequently until the next check for LoH/AP response.

This is a very complex mechanics. On one hand, damage is applied separately from, on the other hand, crit damage seems to be intertwined with the frequency of LoH/AP response ticks.

By the way, you could post a link to my Arcane Torrent mechanics post in your "Skill-Specific Mechanics" section if you like.
http://us.battle.net/d3/en/forum/topic/8728752791

Blizzard
Does not stack. DoT ticks 4 times per second, not affected by attack speed. Every tick can crit individually. Damage displayed once every 0.5 seconds.
Extended duration from gear or via Unrelenting Storm rune does not result in the same numer of ticks spread out over the longer duration. Per second added, the number of ticks is increased by 4 and total damage is increased by 85%


Again, you´d have to differentiate between damage ticks and LoH ticks. Damage is applied each frame, displayed each 0.5 second over monster heads. LoH is triggered 4 times per second.

I´ll post more later this was long enough as my "intro".
Edited by Nubtro#2147 on 6/18/2013 4:48 AM PDT
Reply Quote
nice analysis
Reply Quote
Hey, thanks for dropping by Nubtro :)

06/18/2013 04:37 AMPosted by Nubtro
I guess I won´t have to reasearch Wizard skills with you around anymore ;) although recently I began playing the class more, trying out skills and stuff, it´s quite fun.

There's still enough things we don't know ;)
For example I know you frame-by-framed almost every Barb skill and found most had a base frame length that was NOT 60. But I couldn't bring myself to do the same for Wizard spells. I did a few and saw 60 frames (animation start to animation end). Then again, I recently also had some hickups with either recording or playback (or both), so I guess I should say "somewhat close to 60". Long story short: Wiz casting animations are not properly researched.


Advanced research as a Wiz:
-> low damage white weapon, +min damage jewelry which brings minimum damage higher than maximum (if min>max the game will use min=max as the new base weapon damage so you get steady values)
-> get rid of all life steal and life regen, only use LoH if you want to test LoH
-> get hit (don´t be at max health) when testing LoH -
-> Frost Nova the monster to have a short window for testing without the risk of being killed - note that MP0-1 might be best to not get one shot as you´re not killing stuff with your <150 weapon damage weapon
-> first test should be with a 2H 1.00 aps weapon to get a basic understanding of the skill mechanics
-> put the skill you´re testing on a keyboard button, because the game doesn´t visually display mouse clicks so you wouldn´t be able to determine when the "click" occured
-> the click (skill activation) is usually displayed as a 5 frame long darkening of the skill icon, which you´ll recognize when viewing the video frame by frame
-> if you research LoH, put your mouse cursor on your health globe to record health changes
-> if you research resource regen/skill cost mechanics, the mouse cursor should be on the resource globe - note that in both cases, you may have to adjust your positioning to hit a target in the direction of either globe
-> most recording software displays ingame frame rate, if not use ctrl+r (built in game frame rate) - to have the most accurate results wait until you have a steady 60fps rate, at least for a couple of seconds - if it´s not steady the frame lengths jump around and may make it very difficult to figure out the mechanics
-> record the video and then view it frame by frame
-> if you lack patience, don´t research game mechanics

Thanks, will definitely include that.
Regarding not getting killed: Very first quest of A1. See my hints for theorycrafting, third paragraph. It's not ideal because you can't port back to town to change skills or pick up things from your chest, but other than that it's great.

I believe the best way to describe skills is by separating each aspect of it:
- damage and its breakpoints, if any
- resource cost and its breakpoints, if any
- on attack, on hit and on crit procs and LoH

Yeah, I tried to do that. Maybe I should try to make it more clear.

1. skill damage = aps * 6.0667% (364%/60) weapon damage each frame (aps * 364% / sec)
-> damage is applied each frame, displayed as a damage tick each 30 frames
-> aps directly increases the actual damage (value) not the frequency of how often damage is applied
-> first damage tick occurs 5-6 frames after the skill activation (skill activation is frames 0-5)

[...]

The problem here is the fact that damage has a different scaling than cost and LoH so we don´t know if the game only considers the one frame when it applies the LoH/AP cost tick to determine if APoC is triggered.

By looking at the monster health values ticking each frame I was able to see that crits occur in sets. Note that it´s almost impossible to have a steady framerate to have perfectly accurate results but so far it seems that Sleet Storm crits follow a certain pattern.

Let´s say we´re channeling Sleet Storm at exactly 2.00 aps, which translates to 15 frames per AP/LoH response. The damage is applied each frame, but if the skill crits, you supposedly get 15 frames that crit subsequently until the next check for LoH/AP response.

This is a very complex mechanics. On one hand, damage is applied separately from, on the other hand, crit damage seems to be intertwined with the frequency of LoH/AP response ticks.

That's interesting. I didn't really look at monster health bars yet. But the thing is, a damage tick by itself is boring. We know they are very frequent, DPS is linearly scaling with APS, so breakpoints don't matter.
What's interesting is how they interact with on-crit procs. For on-hit we got this completely unrelated scaling mechanism (LoH/AP cost), but that doesn't apply to on-crit.
With Shocking Aspect or Critical Mass we typically see ike 1.5-3 times as many procs as LoH procs. Sometimes even way more than that. The scaling is weird and hard to analyze due to the randomness of crits.
In addition, the total number of procs is scaling up with APS, so it's either varying tickrates or varying proc rates. That's why I'm a bit hesitant to agree on a fixed tick rate of 60. Might be a display thing only. Or who says it can't be more than 60? (yeah I know, I don't really believe that)

APoC is another uncertainty - proc rates apply to the value of AP you get back, not the frequency. So that's not the same as CM for example.

I agree much of this should somehow be more clearly reflected in my posts, but with all the uncertainties, I can't really figure out a way to better explain it.

Almost the entire "Unsolved Mysteries" section revolves around those issues. There you can also find some links to threads where we tried different things to figure it out. We failed miserably, in part simply because of a lack of patience and motivation.

By the way, you could post a link to my Arcane Torrent mechanics post in your "Skill-Specific Mechanics" section if you like.
http://us.battle.net/d3/en/forum/topic/8728752791

Sure, will do.

Again, you´d have to differentiate between damage ticks and LoH ticks. Damage is applied each frame, displayed each 0.5 second over monster heads. LoH is triggered 4 times per second.

Well, yes, you are of course correct. Damage ticks for Blizzard are probably the same as for the channeled spells like Ray of Frost. But to be honest, I don't think this should be included in this section. Total damage and duration are both fixed, so damage tick rate hardly matters, unless we find a way to definitely link those ticks to on-crit procs. But until then I think the additional info would just cause additional confusion.
Reply Quote
So I´ve spent half a day trying to figure out CM and APoC on channeling skills.

1. Frost Nova deals 1% weapon damage.

2. AP regen
-> gain 1/10 of total regen value each 6 frames
-> has a separate counter that begins the first time you lose AP and it keeps ticking even if you hit max AP

3. APoC with Ray of Frost
-> game applies AP cost of the skill each (floor(30/aps)) frames
-> on the same frame, the game also:
a) heals you for (proc coefficient * total LoH) health
b) checks for a critical hit; if you roll a crit
- all subsequent frames will crit until the next "check" (AP tick)
- you gain (proc coefficient * APoC) arcane power on the frame

4. Critical Mass with Ray of Frost
-> game checks 20 times per second (each 3 frames) for a crit and you have a (100*proc coefficient)% chance to trigger a 1 second CD reduction

I think I solved the following "mysteries":

1. How often do Ray of Frost & Disintegrate actually apply damage?

Each frame after the initiation phase (5-10 frames including activation). The damage per frame is applied to the monster health bar. What the player sees is the damage tick that is displayed above a monster´s head, which is the sum of 30 frames of damage. Note that this "30 frame sum" counter is displayed with a slight delay (not exactly on frame 30 after the initial damage tick).

2. How do Ray of Frost & Disintegrate scale with attack speed?

You deal (aps*tooltip/60) weapon damage each frame. This is unusual, because in case of skills that scale with aps, it is generally used to increase the frequency at which damage is applied. But in this case, aps is a direct damage multiplier.

As a bonus...

3. Deep Cuts DoT
-> is just a true DoT like Barb´s Rend
-> deals (tooltip / (duration * 60)) weapon damage each frame
-> each tick or none crit
-> heals (proc coefficient * total LoH) only once on the initial frame
-> initial DoT frame is the same as the third spectral blade cut so you get the LoH tick twice on that frame for a total of 4 LoH ticks
-> the DoT triggers CM just like other skills

-------------------------------------------------------------------

Hmm, here´s the test I made with Ray of Frost to figure out APoC and CM

1600 intelligence = 17x modifier
20 APoC
10-31 dagger +40-48 source +30 min ring + 30 min ring = 110-110 weapon damage
1.50 aps
50% crit chance
Ray of Frost - Numb vs a single target

110 weapon damage * 17 int bonus * 1.50 aps = 2805 (100% weapon damage)
1 frame normal damage = 2805 * 2.8/60 = 130.9
1 frame crit damage = 130.9 * 3.13 = 409.717

Here´s the exact sequence frame by frame:
monster health - damage - AP globe - what happened
326507 none 145/145 skill activation (skill icon darkened)
326507 none 145/145 icon dark
326507 none 145/145 icon dark
326507 none 145/145 icon dark
326507 none 145/145 icon dark
326507 none 145/145 icon clear
326507 none 145/145 icon clear
326507 none 145/145 icon clear
326098 crit(-409) 143/145(-2) -8 skill cost (16/2) +6.66 APoC proc (20/3)
325688 crit(-410) 143/145
325279 crit 143/145
324869 crit 143/145
324460 crit 143/145
324050 crit 144/145(+1) AP regen (total/10)
323540 crit 144/145
323230 crit 144/145
322820 crit 144/145
322411 crit 144/145
322001 crit 144/145
321592 crit 145/145(+1) AP regen CM proc
321181 crit 145/145
320772 crit 145/145
320362 crit 145/145
319952 crit 145/145
319543 crit 145/145
319133 crit 145/145
318723 crit 145/145
318313 crit 145/145
--------------------------
317904 crit 143/145(-2) -8 skill cost +6.66 APoC proc CM proc(+9)
317494 crit 143/145
317084 crit 143/145
316675 crit 144/145(+1) AP regen
316265 crit 144/144
315855 crit 144/144
315445 crit 144/144
315036 crit 144/144
314626 crit 144/144
314217 crit 145/145(+1) AP regen CM proc(+9)
313806 crit 145/145
313397 crit 145/145
312987 crit 145/145
312578 crit 145/145
312168 crit 145/145
311758 crit 145/145 12292 displayed dmg = 30 tick damage CM proc(+6)
311348 crit 145/145
310938 crit 145/145
310529 crit 145/145
310119 crit 145/145
--------------------------
309988 normal(-131) 137/145 -8 skill cost
309857 normal 138/145 AP regen
5x normal 138/145
309072 normal 139/145 AP regen
5x normal 139/145
308286 normal 140/145 AP regen
308155 normal 140/145
308025 normal 140/145
307894 normal 140/145
307763 normal 140/145
308155 normal 140/145 weird frame jump 3 frames back
308024 normal 140/145
307893 normal 141/145(+1) AP regen
--------------------------
307091 ?crit? 139/145(-2) -8 skill cost +6.66 APoC proc
302585 crit CM proc(+26)
291202 crit 133/145 CM proc(+55)

CRITICAL MASS
+9+9+6+26(27)+55(54)+23(24)+10(9)+9+22(21)+1(3) close to CD end

Continuation of the sequence after the next Diamond Skin cast

220871
218253
0-5 diamond skin click
216682 normal 108/145 skill icon darkened + CD icon appeared
216551 normal 109/145 AP regen
216420 normal 109/145
216289 normal 109/145
216159 normal 109/145
216028 normal 109/145
215897 normal 109/145
215766 normal 110/145 AP regen
-----------
215356 crit 108/145 -8 skill cost +6.66 APoC proc
214947 crit
214537 crit
214127 crit
213718 crit
213308 crit 109/145 AP regen
212898 crit
212488 crit
212079 crit
211669 crit
211259 crit
210850 crit 110/145 AP regen
210440 crit
210030 crit
209620 crit
209211 crit
208801 crit
208392 crit 111/145 AP regen CM proc
207981 crit
207572 crit
-----------
207162 crit 110/145 -8 skill cost +6.66 APoC proc
206752 crit
206343 crit
205933 crit 111/145 AP regen
205523 crit
205113 crit
204704 crit
204294 crit
203885 crit
203475 crit 112/145 AP regen CM proc(+12)
203065 crit
202655 crit
202245 crit
201836 crit

198967 crit 112/145 CM proc(+11)
196100 crit 114/145 CM proc(+7)
195280 crit 114/145 CM proc(+2)
191183 crit 116/145 CM proc(+10)
186267 crit 116/145 CM proc(+12)
183809 crit 117/145 CM proc(+6)
176434 crit 119/145 CM proc(+18)
171517 crit 120/145 CM proc(+12)
170697 crit 120/145 CM proc(+2)
166600 crit 122/145 CM proc(+10)

From this test, it seems as if the CM procs have a common denominator, which is 3 frames. The ones that didn´t can be easily explained by frame length jumping around because of not having a steady framerate between two CM procs (this happens quite often between two subsequent ticks no matter if they´re damage or resource globe response ticks and no matter what skill I test).

Here for example I got 11 frame difference (12-1) followed by 7 frame difference (6+1) followed by 2 frame difference (3-1) followed by 10 frame difference (9+1) before it jumped back to a steadier 12/6/18/12 frame difference between CM procs.

I´d go even as far as to say that Critical Mass starts a unique counter that checks each 3 frames after the initial proc whether you scored a critical hit and how many. I don´t think it separately calls this for each skill you use but overall, which would make it easier to implement/code IMO.

What do you guys think?

So to simplify the result, at 1.50 aps the game performs the following actions each 30/aps = 20 frames:

1. LoH recovery, which means you get a health globe/bar response
2. AP cost, which means you get a AP globe response
3. rolls its RNG for a critical hit and if you crit
-> you gain resource from APoC
-> all 19 subsequent frames until the next "check" will crit

In case of CM, the game checks each 3 frames for a crit. This means that if you didn´t roll a crit on the determining AP response frame, you would get zero CM procs during the subsequent 19 frames (20 total) but if you did roll a crit, you might get 6-7 CM procs withing those 20 critting frames (very unlikely because of the proc coefficient).

EDIT1: To confirm this, it would be cool if a super fast Wicked Wind Wizard with maxed out crit chance made a video and checked it frame by frame to see whether he got a faster CM proc frequency than each 3 frames.

My pinpoint barrier/deep freeze/sleet storm didn´t give me a faster CM proc rate than 3 frames. The problem is the fact that you may get more than 1 CM proc on the same frame so in an extreme case of hitting many mobs with many Winds/Sleet Storm the CD may be lowered so fast that you don´t get to see 2 subsequent CM procs. Maybe here Archon CD could help but it´s almost impossible to spot CM procs on its icon...

EDIT 2: The "3 frame" thing probably only applies to certain skills like Ray of Frost that deal damage each frame. It being a single overall counter doesn´t make sense otherwise stuff like Frost Nova proccing CM would rely on you hittin the "3" in addition to rolling a crit on it.

Speaking of which, Frost Nova may lower its own cooldown lol.
Edited by Nubtro#2147 on 6/18/2013 1:13 PM PDT
Reply Quote
Wow, great job.

But, umh, this isn't scaling with APS, or am I missing something?

Let's say 1 APS, that's
30 frames check interval
-> max 60/30=2 crit intervals per second
-> max 30/3=10 CM procs per crit interval

At 2 APS it's
15 frames check interval
-> max 60/15=4 crit intervals per second
-> max 15/3=5 CM procs per crit interval

Over a fixed duration, those two cases should be quite some procs apart. I mean really significantly.

[edit]Oh, are you sure about 1% for Frost Nova? I am pretty certain that when I first saw it, it was a tad over 1%, more like 1.25%
Edited by apo#2677 on 6/18/2013 12:20 PM PDT
Reply Quote
Wow, great job.

But, umh, this isn't scaling with APS, or am I missing something?

Let's say 1 APS, that's
30 frames check interval
-> max 60/30=2 crit intervals per second
-> max 30/3=10 CM procs per crit interval

At 2 APS it's
15 frames check interval
-> max 60/15=4 crit intervals per second
-> max 15/3=5 CM procs per crit interval

Over a fixed duration, those two cases should be quite some procs apart. I mean really significantly.

[edit]Oh, are you sure about 1% for Frost Nova? I am pretty certain that when I first saw it, it was a tad over 1%, more like 1.25%


Retesting Frost Nova with a higher damage output.

133 base weapon damage * 21.51 int bonus = 2727.83 (100%)
Frost Nova did 27 damage

It´s 1%.

About CM. At 1.00 aps, within one second you may get
0 frames of crits if none of the two intervals crit = 0 CM procs, zero APoC procs
30 frames of crits if one of the two intervals crits = up to 10 CM procs, one APoC proc
60 frames of crits if both of the two intervals crit = up to 20 CM procs, two APoC procs

At 50% CC the probability of those cases happening would be 25%-50%-25%.

At 2.00 aps, it would be 0-15-30-45-60 crits, up to 0-5-10-15-20 CM procs and 0-1-2-3-4 APoC procs with a similar probability which means you´re correct when you say CM doesn´t seem to scale with aps in case of Ray of Frost as it would only depend on your crit chance.

I redid the test with 1.00 aps and even 2.43 aps activated ingame framerate display and when I was at a steady 60fps I mostly got the CM procs separated by a common 3 denominator frame lengths.

I don´t really know how to confirm this theory. But it´s clear from my tests that CM really procs much more often than for example APoC when you use skills like Ray of Frost.
Reply Quote
Yeah finally a sticky! ^_^

And that's some real interesting research you did Nubtro. Thanks for sharing!
Edited by TekkZero#1963 on 6/19/2013 10:48 AM PDT
Reply Quote
Did you guys ever resolve the number of ticks/sec scaling inversely to APS for archon disintegration?
Reply Quote
Holy crap, it actually happened. Thanks, dear mysterious moderator!

@silverfire: We're kinda onto it. Thanks to Nubtros input, I feel like we're really close to finding almost all the answers to all our open questions.

@Nubtro: Thanks for verifying FN damage. Will include this in my next update.
Regarding CM procs, I think there has to be just one more step we haven't quite figured out. Because it really should scale with APS.
We already did some tests with WW, RoF and I think also Disintegrate. The easiest test I think is cast Archon, cancel it, then some time during the cooldown cast a fixed number of WWs or a fixed duration of RoF/Disintegrate. The difference between Archon tooltip cooldown and actual cooldown time (stopwatched) is the total number of CM procs. Of course due to random crits, we have to do this multiple times and take an average.
In the past we already did this for different APS values with keeping all other values constant and the total number definitely scales up. It doesn't seem to be linearly scaling, but higher APS does result in more procs.
I will do some more tests on this.
But as always, just getting your view on things was a huge step forwards. Big thanks for that.
Reply Quote
Grats on the sticky.

Just checking in, but did you guys find the source of weird channeling spell costs when below 1 aps and combined with cost reduction gear?
Reply Quote
Holy crap, it actually happened. Thanks, dear mysterious moderator!

@silverfire: We're kinda onto it. Thanks to Nubtros input, I feel like we're really close to finding almost all the answers to all our open questions.

Forgive me for asking about this narrow topic.

2. How do Ray of Frost & Disintegrate scale with attack speed?

You deal (aps*tooltip/60) weapon damage each frame. This is unusual, because in case of skills that scale with aps, it is generally used to increase the frequency at which damage is applied. But in this case, aps is a direct damage multiplier.

If this is true, doesn't it directly contradict what you found with your actual damage actually decreasing with higher APS?
Reply Quote
Grats on the sticky.

Just checking in, but did you guys find the source of weird channeling spell costs when below 1 aps and combined with cost reduction gear?

Nope, not yet.

06/19/2013 11:57 AMPosted by silverfire
If this is true, doesn't it directly contradict what you found with your actual damage actually decreasing with higher APS?

I don't really know what you mean. Did we find that? I get confused, it's too hot right now, my brain is melting.
I vaguely remember I did some confusing tests which at first pointed in the wrong direction due to insufficient sample size. But in the end, I arrived at the conclusion that DPS linearly scales with APS, which perfectly fits Nubtro's formula as quoted above. According to this, damage is applied at a fixed rate of 60 times a second and damage value (not frequency) is scaling with APS. So that makes DPS completely independent from breakpoints.
Reply Quote
I meant this:

I did this 3 times for each setup. And here's my damage values:
>2 APS: 4471
>2 APS: 4458
>2 APS: 4454
<2 APS: 4529
<2 APS: 4595
<2 APS: 4560

No, I did not get the <> wrong. I really did more damage with my lower breakpoint and lower DPS setup. No idea what the hell this is about.
Reply Quote
Yes, I thought so. I partially revised that later in the same thread:
https://us.battle.net/d3/en/forum/topic/8197583150?page=3#46

Reading it again, I think I did not make that very clear. But the values at that point suggest DPS is simply linearly scaling with APS.
Reply Quote
Jeez...RNG is RNG even when it comes to damage numbers...
Reply Quote
Grats on the sticky! One of these days I'll go an test all the mob resistances, even though they are somewhat inconsequential.
Reply Quote
Congrats to sticky. It would be probably a wise choice to edit out the part where you criticize developers, just to make sure it stays sticky (the person responsible probably didn´t get to that part) haha...

Anyway, I just couldn´t let it be so again today (yesterday) I spent a couple of hours recording just that, Archon CD reduction on a single target with Ray of Frost.

Basically we would need many many tests because at 50% CC that I tested, it´s really just a coinflip whether you get a crit or CM proc...I ranged from 1.5 procs/sec to 2.7 procs/sec within 10 tests :/

Here´s the data. It´s inconclusive. Ideally we would do this with 100% crit chance but that´s impossible with a Wiz as far as I know. You also have to find a mix of MP/gear to be able to channel RoF against a single target for 40 seconds without being killed or without killing the monster. You have to get creative...

First row means start of Ray of Frost in the video (initial damage tick) and the first value in parenthesis is the number of frames since I activated Archon (the reduced duration one), deactivating it and until the initial ray of frost tick. A lot of frame by frame clicks...I´m probably mad for doing this, oh well....

Second row is end of Archon cooldown, the next value is the time it took to reset the cooldown with RoF = how many CM (assuming 60 frame) procs I´ve gotten in how many seconds of channeling it.

I´ve underlined the end result of each set of tests

----------------------------------------------------------------------------------------------
50% CC 1.50 aps Cold Blood 0.333 proc coefficient
----------------------------------------------------------------------------------------------
20 sec 00 frames Archon->RoF (90 frames)
53 sec 15 frames 33 sec 15 frames (1995 frames) = 65.25 procs / 33.25 sec 1.9624/sec
15 sec 20 frames Archon->RoF (81 frames)
55 sec 50 frames 30 sec 30 frames (1830 frames) = 68.15 procs / 30.50 sec 2.2344/sec
8 sec 58 frames Archon->Rof (14-20-20)
35 sec 30 frames 26 sec 52 frames (1612 frames) = 72.90 procs / 26.87 sec 2.7134/sec
16 sec 55 frames Archon->RoF (20 frames)
47 sec 44 frames 30 sec 49 frames (1849 frames) = 68.85 procs / 30.82 sec 2.2342/sec 83 sec 16 frames Archon->RoF (39 frames)
128 sec 5 frames 34 sec 49 frames (2089 frames) = 64.53 procs / 34.82 sec 1.8535/sec 7 sec 16 frames Archon->RoF (43 frames)
39 sec 37 frames 32 sec 21 frames (1941 frames) = 66.93 procs / 32.35 sec 2.0690/sec 105 sec 49 frames Archon->RoF (60 frames)
133 sec 29 frames 27 sec 40 frames (1660 frames) = 71.33 procs / 27.67 sec 2.5783/sec 4 sec 40 frames Archon->RoF (54 frames)
44 sec 13 frames 39 sec 33 frames (2373 frames) = 59.55 procs / 39.55 sec 1.5057/sec
130 sec 8 frames Archon->RoF (47 frames)
163 sec 21 frames 33 sec 13 frames (1993 frames) = 66.00 procs / 33.22 sec 1.9870/sec
5 sec 19 frames Archon->RoF (57 frames)
37 sec 46 frames 32 sec 27 frames (1947 frames) = 66.60 procs / 32.45 sec 2.0524/sec
---------------------------------------------------------------------------------------------- 60000 CD frames 525 CD frames outside of RoF 19289 CD frames with RoF
40186 CM frames = 669.76667 procs / 321.48333 sec 66.98 procs / 32.15 sec 2.0834/sec

----------------------------------------------------------------------------------------------
40% CC 1.50 aps Cold Blood 0.333 proc coefficient
----------------------------------------------------------------------------------------------
7 sec 35 frames (44 frames)
38 sec 1 frame 30 sec 26 frames (1826)
60 sec 58 frames (64 frames)
96 sec 35 frames 35 sec 37 frames (2137)
4 sec 10 frames (46 frames)
41 sec 50 frames 37 sec 40 frames (2260)
48 sec 35 frames (86 frames)
96 sec 2 frames 47 sec 27 frames (2847)
4 sec 33 frames (60 frames)
40 sec 10 frames 35 sec 37 frames (2137)
53 sec 52 frames (53 frames)
91 sec 18 frames 37 sec 26 frames (2246)
2 sec 42 frames (50 frames)
41 sec 53 frames 39 sec 11 frames (2351)
48 sec 53 frames (104 frames)
87 sec 32 frames 38 sec 39 frames (2319)
127 sec 23 frames (45 frames)
162 sec 51 frames 35 sec 28 frames (2128)
169 sec 2 frames (86 frames)
202 sec 23 frames 33 sec 21 frames (2001)
----------------------------------------------------------------------------------------------
60000 CD frames 638 CD frames outside of RoF 22252 CD frames with RoF
37110 CM frames = 618.50 procs / 370.86667 sec 61.85 procs / 37.09 sec 1.6677/sec
----------------------------------------------------------------------------------------------
50% CC 1.20 aps Cold Blood 0.333 proc coefficient
----------------------------------------------------------------------------------------------
4 sec 27 frames (55 frames)
35 sec 51 frames 31 sec 24 frames (1884)
98 sec 23 frames (71 frames)
136 sec 27 frames 38 sec 4 frames (2284)
3 sec 30 frames (62 frames)
38 sec 40 frames 35 sec 10 frames (2110)
4 sec 38 frames (74 frames)
38 sec 40 frames 34 sec 2 frames (2042)
3 sec 42 frames (46 frames)
45 sec 11 frames 41 sec 29 frames (2489)
73 sec 43 frames (53 frames)
105 sec 56 frames 32 sec 13 frames (1933)
109 sec 2 frames (49 frames)
138 sec 28 frames 29 sec 26 frames (1766)
3 sec 41 frames (65 frames)
41 sec 22 frames 37 sec 41 frames (2261)
4 sec 6 frames (44 frames)
37 sec 3 frames 32 sec 57 frames (1977)
44 sec 33 frames (61+55 frames)
78 sec 2 frames 32 sec 34 frames (1954)
----------------------------------------------------------------------------------------------
65.32 / 34.5 1.8932/sec
7 sec 2 frames (63 frames)
43 sec 0 frames 35 sec 58 frames (2158)
47 sec 23 frames (47 frames)
88 sec 25 frames 41 sec 2 frames (2462)
----------------------------------------------------------------------------------------------
72000 CD frames 745 CD frames outside of RoF 25320 CD frames with RoF
45935 CM frames = 765.58333 procs / 422 sec 63.80 procs / 35.17 sec 1.8142/sec
Edited by Nubtro#2147 on 6/19/2013 5:02 PM PDT
Reply Quote
Nubtro:
Here´s the data. It´s inconclusive. Ideally we would do this with 100% crit chance but that´s impossible with a Wiz as far as I know. You also have to find a mix of MP/gear to be able to channel RoF against a single target for 40 seconds without being killed or without killing the monster. You have to get creative...
You can get +100%cc with Ray of Frost, but it requires:
*High critical chance - like 60%. So you'd use things like Nat's Ring+Boots with the ring having some intrinsic cc)
*Pickup some gear with %Crit Chance for Ray of Frost skill (ie Wizard Hat, Maras, and Skull Grasp)
*Use Energy Armor (Pinpoint Barrier)
*Trigger the +15%cc from Frost Nova (Deep Freeze) - This one is the issue I think, since the bonus doesn't last longer than 12s.
*3%cc bonus from Scoundrel.

People have done 100%cc before with Energy Twister, so it should be basically the same principle (albeit still a PITA). :) I believe you can get ~82%cc for Ray of Frost just from gear. So that's about 90% without resorting to Frost Nova (Deep Freeze).
Reply Quote
'Grats on the sticky :)
Reply Quote
wow gratz, finally
Reply Quote

Please report any Code of Conduct violations, including:

Threats of violence. We take these seriously and will alert the proper authorities.

Posts containing personal information about other players. This includes physical addresses, e-mail addresses, phone numbers, and inappropriate photos and/or videos.

Harassing or discriminatory language. This will not be tolerated.

Forums Code of Conduct

Report Post # written by

Reason
Explain (256 characters max)
Submit Cancel

Reported!

[Close]