Heroic Megaera’s Diffusion Mystery

EDIT: I updated the numbers in the post following the excellent comments by Vixsin and Lakh. I have very little background with this sort of thing, so the help is very appreciated! I hope no one read the post expecting revelations and great number gymnastics. I’m only a curious healer who tends to fixate on minor details and who doesn’t turn away from a challenge. I have no answers myself, but there are some great comments in the comment section for those looking for information.

Ever since I stumbled across a parse where I randomly did 5 million healing from Diffusion, I’ve been fixated on figuring out how that spell works.

magsspellsbefore

Sadly, after hours and hours of pulling my hair out, I still don’t know.

In theory, Diffusion (available only on Heroic mode) is a debuff the Arcane Head puts on anyone who gets hit by its Arcany breath. The debuff redirects 10% of healing received to targets within 8 yards. It also stacks, which, if I understand right, would redirect 30% of healing received.

In practice, it is certainly a debuff put on Breath takers by the Arcane head that causes nearby players (and pets) to get healed, but how exactly it works is a still a mystery to me, even after I’ve gone nearly bald going through logs.

Anyone who, following this post, has further insight, is welcomed to share it. I’m happy to provide links to the actual logs if anyone wants to give it a go for themselves as well. Maybe I’ll finally get some closure and be able to move on with my WoW life.

Everything under the cut. Also, if you’re looking for guides on working World of Logs, I suggest you check out this post and this post, since I won’t be going into details on the how-tos.

Who Had Stack of Diffusion?

DiffusionTargets

As you can see here, both our tanks and only our tanks were affected by the Diffusion debuff.

A Look at Our First Diffusion

diffusiontheck

For visual types, here is the uptime of Theck’s Diffusion. Let’s select approximately that period of time for the Log Browser.

selectiont

Aaaand the queries used.

querytheck

Theck was my Beacon target and I wasn’t sure if Beacon could trigger diffusion so I included it. I realized as I was doing this writeup after hours of taking screenshots of my logs that I probably should have included Arcing Light, Holy Radiance and Daybreak. I really don’t have it in me to redo all my shots (I’m embarrassed enough about the amount of time I’ve wasted on this one, useless detail of a spell) and these heals are unlikely to be relevant, but I’ll add their lines in by hand. And if anyone thinks I’m making up the weirdness of Diffusion, they’re welcome to have a look at my logs themselves.

So for the first tick of Diffusion.

diffusion1t

Beast got a nice overheal of 19556.

Since it was shortly after the third stack of Diffusion was applied and we can leave some room for delay, we should consider that the heal could have procced off two stacks, or off three. The heals needed to proc a Diffusion heal of 19556 are:

If 3 stacks (30%): 65187 – 19556 (absorbed) = 45631
If 2 stacks (20%): 97780 – 19556 (absorbed) = 78224

Or if Beast’s Hunter is glyphed, the Diffusion heal should have actually been (19556 * 0.9) = 17600

Meaning

If 3 stacks (30%): 58667 – 17600 (absorbed) = 41067
If 2 stacks (20%): 88000 – 17600 (absorbed) = 70400

Even assuming that Beacon and overheals can trigger Diffusion, the only value in that ballpark is a 40k overheal at 21:49:40.190.

Let’s look at the next tick of Diffusion.

diffusion2t

The screenshot got cut off because I was only interested in the Diffusion value, but I should probably add that most of the raid was healed for the same thing as me, 22874. (Yay for Diffusion during Rampage!)

Also missing on the screenshot (after adding Daybreak, Arcing Light and Holy Radiance) are:
[21:49:59.476] Rykga Arcing Light Theck +3242
[21:50:01.550] Rykga Arcing Light Theck +3291
[21:50:01.550] Rykga Holy Radiance Theck +3303

For a 22874 tick at 3 stacks of Diffusion (30%), we would need 53373 (which is 76247 – 22874) healing.

The only thing that would come somewhat near that is:
[21:49:45.859] Rykga Beacon of Light Theck +41980 (O: 43649)

But, again, this is assuming Beacon triggers Diffusion. And that a whole 15 seconds delay on the tick is normal. And that we’re off by 12k.

Another theory would be that healing done accumulates on a target and is released at once later on (after all, the Diffusion description reads “all healing“). We can look into that, but let’s choose our other tank, the one who isn’t getting 23948723894 heals from Beacon of Light.

Diffusion #2

diffusionagwyne

This is the uptime from our tank who took Diffusion stacks later on in the fight. Let’s select this part of the fight for our Browsing pleasure.

selectiona

And the first tick!

diffusion1a

There were no extra heals from Arcing Light/Daybreak/Holy Radiance.

One tick of Diffusion on Beast for 13692. Agwyne clearly had only one stack of the debuff (10%), meaning that we would have needed a 123228 (136920 – 13692) heal to trigger this.

Or if Beast’s Hunter is glyphed, then the Diffusion tick was originally 12321, meaning it would have needed 110889 (123210 – 12321).

Lay on Hand was too powerful and nothing else even came close. I calculated the cumulative healing between the Diffusion application and the Diffusion trigger, giving me a total of 94416, excluding Lay on Hands, a number that isn’t quite there. So that theory doesn’t work either.

Onto the next tick.

diffusiona2

Again, it’s unclear if the tick on Beast came from 1 or 2 stacks of Diffusion, so considering either possibility, to get a heal of 10554/9499, we would have needed 94986/85491 for one stack (10%) or 42216/37996 (20%).

There was one Eternal Flame for 82305, which, I guess, is pretty close, but there is a 10 second delay.

One final chance!

diffusiona3

Including the extra spells, we’ve got:
[21:53:43.792] Agwyne afflicted by Diffusion (3) from Arcane Head
[21:53:44.134] Rykga Holy Shock Agwyne +*78504*
[21:53:45.790] Rykga Divine Light Agwyne +*140007* (O: 744)
[21:53:45.856] Rykga Eternal Flame Agwyne +0 (O: 7316)
[21:53:48.235] Rykga Eternal Flame Agwyne +*15071*
[21:53:50.731] Rykga Eternal Flame Agwyne +0 (O: 7316)
[21:53:51.466] Rykga Holy Shock Agwyne +*0* (O: 78211)
[21:53:54.714] Rykga Holy Radiance Agwyne +0 (O: 17016)
[21:53:55.105] Rykga Eternal Flame Agwyne +*0* (O: 134561)
[21:53:57.153] Rykga Eternal Flame Agwyne +0 (O: 7316)
[21:53:57.889] Rykga Arcing Light Agwyne +*0* (O: 11355)
[21:53:59.235] Rykga Eternal Flame Agwyne +7316
[21:53:59.969] Rykga Arcing Light Agwyne +3551
[21:54:01.279] Rykga Holy Radiance Agwyne +3630
[21:54:01.340] Rykga Eternal Flame Agwyne +7316
[21:54:01.855] Rykga Arcing Light Agwyne +3493
[21:54:01.964] Rykga Diffusion Rykga +39217
[21:54:02.307] Rykga Diffusion Astart +39217

For a Diffusion heal of 39217 at 3 stacks, we’d need a trigger heal of 91506.

The Eternal Flame at 21:53:55.105 (134561) overshoots by 40k and is all overheal. There was an effective heal of 140007, but that was a good 16 seconds before the actual proc.

Alternatively, if you add up the preceding heals, you get 40377 between the Diffusion and 21:53:48.235.

Conclusion

There doesn’t seem to be an obvious way Diffusion works.

It doesn’t only trigger off big heals since none of my large Diffusion heals were triggered by a single heal. Nor do the heals from a single healer add up over a few seconds to be released at once.

The only other theory I could think of (other than the Diffusion heals being completely at the mercy of an RNG with a sense of humour) is that all heals (from all healers) on a Diffusion tank accumulate for a few seconds, then are released and attributed either randomly to a participating healing, or to the last healer to contribute to the Diffusion heal pool. (If you remember the odd healing attribution from Judgement of Light from back in the day, that is what this reminds me of.)

But as curious enough as I am, I’m not quite motivated enough (at this time) to investigate that theory.

Regardless, my statement from the Megaera logs post still stands:”Because Diffusion seems to only heal when it feels like it, without involving much control from the player, when analyzing logs and comparing my performances, I’d subtract Diffusion healing from my total healing count.

Oh, and last raid week I did try tracking Diffusion on my frames to squeeze in as much healing on the tank as possible. What to see what happened?

(Note, I totally forgot to swap from Holy Prism to Light’s Hammer again. *fail* This is why I can’t have nice things.)

healingdone

Not even a million! Diffusion works better when you’re not trying.

About these ads
Explore posts in the same categories: Boss Fight: Pandaria, World of Logs

Tags: , , , , , ,

Both comments and pings are currently closed.

16 Comments on “Heroic Megaera’s Diffusion Mystery”

  1. Lakh Says:

    There’s a glyph that increases the healing done to hunter pets by 20%, I dunno if that would apply to random summon pets or not, but if so that’d put the initial heal around the 8k mark in that example.

    I’d rather assumed that if someone wasn’t within 8 yards (of the head? of the tank?), the heals just got pissed away into the nether. But that isn’t terribly scientific. Still got 1-2 other bosses to go before we’ll start work on this tho, so I hadn’t examined it closely yet.

    • Lakh Says:

      Getting it mixed around – Glyph of Animal Bond & only 10%. So still not on the money for an explanation.

      • Ophelie Says:

        Darn Beast. Why did he have to be the one to soak all the Diffusions, haha. And of course, he never took a Diffusion heal along with anyone else, so I don’t know if his hunter was glyphed. I noticed the other Hunter pet was getting an extra 10%, so I redid all the numbers to account for the possibility that Beast might be getting bonus healing.

        Thank you!

  2. Vixsin Says:

    Well, I can help unravel part of the mystery for you through my own logs by looking at Riptide ticks on the tank before and after Diffusion is applied.

    Before Diffusion:
    [20:35:57.978] Vixsin Riptide Spoodzn +0 (O: 11715)

    But when Diffusion is applied:
    [20:37:48.444] Spoodzn afflicted by Diffusion from Arcane Head
    [20:37:49.260] Vixsin Riptide Spoodzn +0 (O: 10543)

    And when you remember the wording on Diffusion, that it absorbs 10% of healing done:

    11715 x 0.9 = 10,543.5

    11,715 = amount of an RT tick before Diffusion
    10,543 = amount of an RT tick after Diffusion

    So, the assumption based on the above would be that Diffusion absorbed 10% of the amount that RT would have ticked for.

    It’s also interesting to note, when looking at Diffusion’s behavior in logs, that it looks and acts very similar to Resto Shamans’ Ascendance, which aggregates healing from a prior segment of time and then distributes in equal portions to the surrounding players at ~1sec intervals. In that regard, it’s also interesting to note that the Diffusion heal is not affected by Mastery, cannot Crit, and is subject to class healing modifiers (+10% for hunters and locks, +20% to feral druids, which is why if you look at the Diffusion heal on any of those classes, it will be more than other affected players receive).

    Not much help, I know, but at least it’s something?

    • Ophelie Says:

      Ooo you are right, I totally forgot about the part where the Diffusion heal takes away 10% (20%, 30%) of the heal the tank would have taken.

      This is why I don’t typically do numbers posts, haha. I always forget something basic and important. I am good at finding information, not so much at processing it. I adjusted the post to take into account the healing “absorbed” by the debuff.

      I wasn’t familiar with Ascendance until now (I admit that I’m very weak when it comes to understanding other healing classes), but the way it aggregates healing then redistributes it makes sense and make it easier for me to believe they would have used a similar mechanic for the raid. The differences would be that Diffusion seems to spit out all it’s healing at once instead of at intervals, and takes in the sum of everyone’s healing, not just one person.

      It seems to go either by time or by quantity of healing accumulated. When the tank was off in his corner getting his regular tank heals, it took awhile for Diffusion to go off, and then it seemed to go off inconsistently (but perhaps that was the result of only beast being there). But during Rampage, it went off every 1-2 seconds

      Just because I want to post them somewhere, this is a summery of the first Diffusion used in this post.

      [EDIT: I suppose to get a clear view of it, I should look at Diffusion for all healers, not just me... My brain is kind of frazzled by now.]

      [21:49:23.642] A Diffusion Beast +0 (O: 9156)
      [21:49:24.844] T Diffusion Beast +0 (O: 15729)
      [21:49:25.687] S Diffusion Beast +0 (O: 27682)
      [21:49:26.850] S Diffusion Beast +0 (O: 30199)
      [21:49:27.714] F Diffusion Beast +0 (O: 18162)
      [21:49:37.792] T Diffusion Beast +0 (O: 26798)
      [21:49:40.662] At Diffusion Beast +694 (O: 7749)
      [21:49:41.834] T Diffusion Beast +0 (O: 20421)
      [21:49:43.829] R Diffusion Beast +0 (O: 19556)
      [21:49:58.867] S Diffusion Meeperz +12156
      [21:49:58.867] S Diffusion Cuponoodles +0 (O: 13371)
      [21:49:58.867] [8 more ~ 12156 Diffusion Heals from S]
      [21:50:00.778] H Diffusion Hyeri +19866
      [21:50:00.942] [37 more ~ 22875 Diffusion Heals from H]
      [21:50:01.740] Rykga Diffusion Rykga +22874
      [21:50:02.081] Rykga Diffusion Astart +22874
      [21:50:02.081] [~30 more Diffusion Heals from Rykga]
      And so on every 1-2 seconds.

  3. Nzall Says:

    If you’re running with a shaman, it’s easy to see that Diffusion is just added to the last person to cast, because it’s added to SLT as well.

    if you want to get techy, go to the analysis Pane. I’m using the current top holy paladin log from Envy-Kazzak for analysis. Diffusion did a total of 2.72% healing from EVERY healer. I think if you do some math on how high diffusion healing % is on average from all debuffed targets, it’ll be close to that number. especially if it’s spread evenly across the raid.

    • Ophelie Says:

      What is your 2.72% in reference too? Do you mean Diffusion healing accounts for 2.72% of the healing each healer did while the Diffusion debuff was active?

      I’m not sure what you mean by “last person to cast”. Is that assuming the healing banking theory? And the person to set off Diffusion once the heal bank is full gets credit? (It does makes sense.)

      • Nzall Says:

        http://www.worldoflogs.com/reports/rt-d6udelwb93rj21fa/analyze/hd/spell/?s=3685&e=4077 is the analysis of healing done divided by spell. if you hover over the diffusion monitor, you’ll see that it did 2.72% of ALL player healing on that attempt for the entire attempt, and that some of it was done by a spirit link totem. This means that WoL just attributes it to random healing.

        as for the 2.72%. Dex had 1 stack 1 time. Javil built up to 3 stacks and then they faded. so Dex has 10% redirect for 45 seconds of a 391 second fight. Javil had 10% up for 17 seconds, 20% for 17 seconds and 30% for 45 seconds. so the 10% was up for 15% of the time, the 20% for 4% of the time and the 30% for 11.5% of the time. however, they both were in rampages, so that means that you take relatively more healing. I think if you calculate the total rampage damage which is dealt to the debuffed tank and took 10% off of that, you’ll end up close to the total diffusion healing across all healers (i.e. that 2.72%). but that would require a lot of math.

        • Ophelie Says:

          I’m not sure I understand the usefulness of knowing that Diffusion did 2.72% overall, even considering stack uptime, since the damage and healing is variable throughout the fight. We’d only be interested in the healing done during Diffusion uptime. I also don’t know what you mean by “WoL just attributes it to random healing”.

          According to the description of Diffusion, Diffusion redirects 30% of the healing done on the Diffusion target. So, according to this description, if Spirit Link totem did 100 healing toward the target, the target would receive 70 and everyone would would either get a heal of 30, or of 30/#oftargets (as the description is unclear). Somewhat similar to a paladin’s Beacon, but with an absorb on the initial target.

          The entire post is about how the description does not reflect what Diffusion actually does. If it behaved like Beacon, the combat log would show way, like waaaaay more ticks. And while we have an idea of how Diffusion works, we still don’t know for sure.

          * * *

          Out of curiosity, I narrowed my sample to the time where Theck had 3 stacks of Diffusion. Over this time, he took 1897037 effective healing. Assuming that represents 70% of the healing he should have taken and that Diffusion would be the other 30% : 1897037*0.3/0.7 = 813015.

          In reality, during that time period Diffusion did 14 990 806 (including overheal).

          This lets us conclude that the 30% healing that goes out is not divided among players (though we suspected that already).

          I have to leave for work so I can’t do this right now, but in theory, if we added each individual trigger of Diffusion, we would hope to get something close to 813015.

          Example:
          If 1 person got 37k
          Then 20 people got 32k
          Then 14 people got 12k

          We’d go 37k + 32k + 12k = the Diffusion calculated from tank healing.

          1- If this is right, then Diffusion most likely works like a heal bank, once the bank is full, it releases a Diffusion tick. Everyone in 8 yards receives the total Diffusion tick.

          2- It is unclear if the Bank is totally heal dependent or if there is a time component too. Since Diffusion ticks are variable, there seems to be both a heal and time component. Or perhaps a randomness component as well.

          3- It’s also unclear which of the healers contributing to the bank receives credit for the Diffusion (though from what commenters are saying, it seems like it is the healer contributing the triggering heal to the bank who gets credit)

          • Calidyn Says:

            I think that’s about how it works. I was poking around your Magaera kill logs and there are some instances where the person who heals with Diffusion hasn’t cast nearly enough heals on the tank -counting from the moment they got debuff- to justify the magnitude of the Diffusion heal.

            I don’t think there is a heal limit. The amount for which it heals is too inconsistent. I think it ticks about every second or so, releasing all the healing it’s gathered since its last tick; and if there isn’t anyone near the tank, the healing is just lost and doesn’t show up on the logs. Since Rampage is the only time the tank has people around him consistently, it’s the only time of the fight where it heals as often as it can, and looking at the logs, this seems to confirm it heals roughly every second. If you tally the heals in the last second, or since the last Diffusion, the numbers seem to fit (also roughly; I’m not sure how to account for inconsistencies – it could be this theory is just wrong some spells not counting for some reason and/or the server disagreeing with the combat log uploaded on the order of certain events, it could also be me doing some incorrect math.)

            The person who gets credit for the heal usually seems to be the last person to land a heal on the debuffed person (regardless of overheal) before Diffusion triggers, though I’ve found some exceptions, where it’s the second-to-last (and one where it was the third to last).

            For the record, this is what I was looking at: http://www.worldoflogs.com/reports/rt-e7f1r6vmzza74tkb/xe/?s=5294&e=5356&x=%28spell+%3D+%22Diffusion%22+and+type+in+%28TYPE_AURA%2C+TYPE_HEAL%29%29+or+%28type+%3D+TYPE_HEAL+and+targetName+%3D+%22Theck%22%29

          • Ophelie Says:

            In one swoop you both uncover the secrets of Diffusion and showcase the usefulness of the dreaded Expression Editor!

            I counted a few Diffusion ticks based on heals and none of them line up, but there’s probably some delay between the heals and the transfer. The heals that went into a particular Diffusion might be 1 or 2 ticks behind and it might not be clear which heals went into which Diffusion.

            I was unsure if the heals ticked on a regular time because there was always a hunter pet next to the tank, but I guess the hunter pet isn’t stationary and might move in and out of the 8 yard radius. On the logs you were looking at, though­, even during Rampage sometimes Diffusion ticked every second, but between 38.500 and 42.412, there’s almost 4 seconds with nothing, even during Rampage. I wonder why that is. The individual heals from Diffusion are also way tinier in that log than in the log I used to write the post, but perhaps we had more damage reduction CDs rolling on that Rampage reducing the amount of effective healing Theck was getting.

  4. Dedralie Says:

    Nice to see someone attacking this! I had given it a brief analysis myself but ultimately had no luck and got distracted by writing up healing behaviour for Twin Consorts and never came back to it.

    One thing I am certain about is that overhealing does not get converted into meaningful healing by Diffusion. Shaman healing is a little easier to analyse – or perhaps I just got a lucky log – but I had a very clear example of a Diffusion heal being exactly 40% (we usually saw 4 stacks of it by the 6th Rampage) of the *effective* healing done by one of my Healing Surge casts. On my phone right now or I’d try to find it for you.

    Because it doesn’t count overheal, it made sense for some – on early, undergeared kills – to have a second player get the Diffusion debuff, so healers could split their attention and not just overhealbomb the same target. We had a few attempts where the healers were so focused on trying to maximise Diffusion throughput on the single Diffusion target, they all ended up producing less Diffusion and less AoE healing in general, and we didn’t survive the Rampage! I think it is good to figure out how it works, but in practice it should be treated as a mechanic that just lets you use fewer raid cooldowns for the affected Rampages.

  5. Dedralie Says:

    I’m so glad to see someone trying to work this out! I noticed massive Diffusion fluctuation between attempts before our first kill, and tried to put some effort into figuring out what was going on, but ultimately we killed it before I got anywhere with the math, and I kinda forgot to ever come back to it :-(

    One thing I can state for certain is that Diffusion does not convert overhealing into raid healing. I have a log where the Diffusion heal I procced with my first cast on the target (whom I was not in range of until Rampage began) was exactly the expected percentage of the *effective* healing dealt by my Healing Surge, In other words it healed for 100k, but 30k of that was overheal, and the Diffusion proc was around 21k, or 0.3x70k. (I am on my phone right now so I can’t find the log but these numbers were close!)

    Before we realised this we had a few attempts where we were all focusing so hard on getting maximised Diffusion healing by overhealbombing our tank, that we produced very little Diffusion healing at all! And since we were spamming him with single-target heals, this meant the rest of the raid suffered, and we didn’t survive the Rampage.

    Because of this it made sense on early kills to get another player – ideally one with a passive healing taken boost, like Hunter or Feral Druid – to accumulate Diffusion stacks as well. This let healers split our healbombing up and reduced the likelihood of overhealing, thus increasing our overall Diffusion output. This really isn’t necessary anymore but it can be helpful for the first few kills!

    I think it is important to understand how Diffusion works (or doesn’t, in the case of Disc Priests, it would seem), but ultimately it is probably best just to consider it like another raid cooldown. It lets you use fewer cooldowns for the 3rd Rampage and it supplements the ones you have for that truly ridiculous 6th Rampage, but the mechanic alone can’t replace real AoE raid healing IMO. :-)

    • Ophelie Says:

      Yeah, I didn’t think Overhealing counted, and I had a discussion with Derevka on Twitter about that some time back, but I wanted to go into the post by just looking at the logs.

      As far as I know, my guild totally ignored Diffusion. The first time I even heard of it (our raid leader prefers when we don’t pre-read strats, he finds it keeps us from being open minded and following his instructions) was when I went through the logs a few weeks ago. I was all “oh, Diffusion, I healed for so much with it, I wonder what it is“. So we never had anyone but the tanks take Diffusion and we never tried to make it tick for more. Rampage wasn’t much of an issue for us (upside to running with 5-6 paladins, I guess. So. Much. Aura. Mastery.) anyway. On our early kills (and still somewhat today) it was always about running out of space due to ice.

  6. Dedralie Says:

    Oops sorry for the double comment. Silly phone! /facepalm

  7. Calidyn Says:

    It won’t let me reply to your reply. ):

    But, yeah, I’m kind of leaning towards timing errors and delays at least contributing somewhat to the discrepancies, mostly because if you look carefully at the timestamps for certain Diffusion heals you’ll see one heal (usually the one on the healer to whom the Diffusion is attributed) several milliseconds before the rest of them.

    I somehow missed the gap while I was looking at the logs originally. I think I didn’t catch all the actual effective heals amongst all the overheals there (which is why trying to solve log mysteries on a couple of hours of sleep is never a good idea, fascinating though they may be). But taking into account that everything from your Diffusion to 39.821 is overheal and then the first actual tick of Diffusion actually happens at 42.192, it’s really just one really MIA tick (the one fitting around second 41).

    It’s still kind of weird, though. The only explanation I can come up with is that Theck stepped just out of range for some reason when the tick was supposed to tick (but then the heals fall short). I’m not familiar with more than the basics for the heroic version of Magaera, so I can’t figure out why he’d do anything but stay put. Plus, the fact that he got two ticks of healing rain during that second would indicate he was definitely within eight yards of the rest raid.

    I don’t know. The whole thing might just be wrong, haha, but if it is – I give up, because I can’t see any other pattern that more or less fits. I think I’m gonna go looking at some 10-man logs tomorrow; it might be easier to find what counts and when if there’s only three healers’ heals to sift though. And if that doesn’t work, maybe in a couple of patches, when I overgear the content and dying horribly is less of a concern, I’ll convince a tank and some more healers to come with me and experiment.

    (Also, to be fair, I could’ve used the log browser to get the same thing; just a lot less of it. I went with the expression editor ’cause it’s what I usually use.)


Comments are closed.


Follow

Get every new post delivered to your Inbox.

Join 1,932 other followers

%d bloggers like this: