Enmity testing - Still Going!

[Hanyoko]

1 Vana'Diel day is exact 3456 RL seconds. That'll make each game minute tick 2.4 seconds, which follows your rounded tick pattern.

Yeah, except as I mentioned before, the ticks are not equal in actual length. They are not 2.4 seconds each, the are in fact either 2 or 3 seconds. The 5 tick pattern is 12 seconds total and repeats itself exactly 288 times per in game day (24 hours * 60 ticks)/(pattern of 5 ticks) = 288. Also, 288*12 = 3456 seconds.

Its probabaly due to round, floor, or ceil used in their calculations.

[Valyana]

Foe Sirvente does not "alter" your enmity, but lowers the decay rate from roughly -60VE/sec to somewhere in the realm of 54-56 VE/sec (I'm just guessing it brings it to -55 VE/sec but I can't prove it at this point). Anyone who has a level 3 Foe Sirvente I would really appreciate help on it.

Since L1 Dirge is -10 Enmity, I'd guess that L1 Sirvente is -10% VE decay rate, or 54/sec, and that 5% per additional merit means L3 Sirvente would be -20%, or 48/sec, which would mean that VE lasts 25% longer. But that's just speculation at this point.

I have L3 Sirvente, and I'll try to test this later tonight.

[Valyana]

I have L3 Sirvente, and I'll try to test this later tonight.

I pull a Bumblebee with my BRD, RDM #2 cures me for 0 twice and I zone.
RDM #1 uses Provoke, which pulls hate for 30 sec.
I zone back in, Sirvente RDM #1 and zone out again.
RDM #2 cures himself for 0.
RDM #1 uses Provoke, which pulls hate for...30 sec.

To the best of my stopwatching abilities, my L3 Sirvente is making no difference at all to how long Provoke holds hate. Which is...surprising.

[Kaeko]

I have L3 Sirvente, and I'll try to test this later tonight.

I pull a Bumblebee with my BRD, RDM #2 cures me for 0 twice and I zone.
RDM #1 uses Provoke, which pulls hate for 30 sec.
I zone back in, Sirvente RDM #1 and zone out again.
RDM #2 cures himself for 0.
RDM #1 uses Provoke, which pulls hate for...30 sec.

To the best of my stopwatching abilities, my L3 Sirvente is making no difference at all to how long Provoke holds hate. Which is...surprising.

Hm this is interesting... We got ours to hold hate for about 33-34 seconds on a Voke with a level 1 Sirvente. Not sure what to say - possibly a bug?

[Ashira]

Try again, except verify neither RDM has Enmity merits, and make sure no one is wearing any Enmity gear. (I know, I know, but have to start at the very bottom basic etc, and doesn't seem like they would be if the voke was 30s, but have to be sure).

Pull your bee, and then have each RDM cast Dispel consecutively, and you do not need to zone.

Then first RDM Provoke to turn it, and time it or whatever if you wish. Then do Sirvente, other RDM Provoke, and see what happens.

If you use the windower, try plugging in recast and use that to time down, as that will give you the 30s mark; or ask the RDM to watch the recast on the ability menu timer.. if it lasts longer than 30, obviously, you have to figure elsewhere from that point.

If it does not last any longer, we'll need to have other Sirvente 3 BRDs test and retest and document, because it it could be another Cleric's Panataloons +1... opening it it works, but the upgrades don't. :/

[Valyana]

While I was at it, I did some quick stopwatch VE approximations for other spells.

Paralyze (and II), Slow (and II), Silence, Refresh, Blink, Aquaveil, Stoneskin, Haste, Phalanx, Regen, Barvirus, Enfire, Poison II all seem to have VE around 300-320 (and 1 CE).

Gravity and Invisible pulled hate very briefly, I'm guessing around 80 VE. Protect 2-4 seem consistent with Protect 1.

Bind is wierd. One time it pulled hate for ~47 sec, another time only about half that. It seemed to hold hate as long as the mob was bound, and for a few swings afterwards. Both RDMs were in melee range the whole time.

Edit: Both RDMs had -4 Enmity in merits and were wearing Macuahuitl +1 to cancel out the merits. I'll try again tomorrow.

[Kaeko]

While I was at it, I did some quick stopwatch VE approximations for other spells.

Paralyze (and II), Slow (and II), Silence, Refresh, Blink, Aquaveil, Stoneskin, Haste, Phalanx, Regen, Barvirus, Enfire, Poison II all seem to have VE around 300-320 (and 1 CE).

Gravity and Invisible pulled hate very briefly, I'm guessing around 80 VE. Protect 2-4 seem consistent with Protect 1.

Bind is wierd. One time it pulled hate for ~47 sec, another time only about half that. It seemed to hold hate as long as the mob was bound, and for a few swings afterwards. Both RDMs were in melee range the whole time.

Edit: Both RDMs had -4 Enmity in merits and were wearing Macuahuitl +1 to cancel out the merits. I'll try again tomorrow.

Thanks for the other numbers. There seems to be a pattern with how SE chose to use CE and VE numbers...

Actions on Enemy - increments of 80, 160, 320, 640, 1280, ...
Abilities on Self - incredments of 100, 200, 300, ...

While not completely thorough, I think we can use this to our advantage when trying to tell things apart. We personally tend to have issues telling 300 apart from 320.

Can anyone just pick one off the list and try to verify if it's 300 or 320? It may help find a better pattern.

[rithridan]

Thanks for the other numbers. There seems to be a pattern with how SE chose to use CE and VE numbers...

Actions on Enemy - increments of 80, 160, 320, 640, 1280, ...
Abilities on Self - incredments of 100, 200, 300, ...

While not completely thorough, I think we can use this to our advantage when trying to tell things apart. We personally tend to have issues telling 300 apart from 320.

Can anyone just pick one off the list and try to verify if it's 300 or 320? It may help find a better pattern.

That's not entirely true, or just has exceptions. Provoke is an action on an enemy yet (supposedly) generates 1800 VE which isn't divisible by 80.

[Kaeko]

Thanks for the other numbers. There seems to be a pattern with how SE chose to use CE and VE numbers...

Actions on Enemy - increments of 80, 160, 320, 640, 1280, ...
Abilities on Self - incredments of 100, 200, 300, ...

While not completely thorough, I think we can use this to our advantage when trying to tell things apart. We personally tend to have issues telling 300 apart from 320.

Can anyone just pick one off the list and try to verify if it's 300 or 320? It may help find a better pattern.

That's not entirely true, or just has exceptions. Provoke is an action on an enemy yet (supposedly) generates 1800 VE which isn't divisible by 80.

That's why I'd like to find some type of pattern. Berserk/Defender seem to be off as well. I don't think so many abilities seem to fall in the 300 or 320 range by chance. I think an accuracy on the VE value within +/- 10 VE is acceptable so this is a bit minor to me personally. I know some people want exact numbers, but it's really hard to test VE given experimental limitations.

[Freysi]

Kaeko, when you did the test where Ashira /kneeled and the mob would turn towards her and then back... did you ever note the frequency of the turns? This might be an important clue to the minimal unit of VE. The enmity tick (however long it is, 1, 1.2, 2,4 seconds perhaps) is imo fundamental to understanding exactly how VE decays. -60 VE per second is a great start, but is it continuous or discrete? If it is continuous that would imply there is a counter on each character ticking down 1 VE at a time. Personally I think it is discrete. Picture the following:

Each player has his enmity counter. At every "enmity tick" the mob simply goes to each player in his list and "eats" an X amount of VE from that list. At the moment it manifests itself as -60 per second, but if the "enmity tick" is perhaps 1.2 seconds that would mean the mob eats a chunk of 72 VE per tick. Also, viewing decay as discrete would explain how identical actions beginning at different times between ticks would still have the identical VE and wear off at the same tick.

For example, say the tick is 1 second. Mob checks his list at 0.0 and sees equal enmity on 2 players. Player A performs an action at 0.1, player B performs an action at 0.9 What exactly happens here? Should B get immediate hate? Has A's VE decayed between 0.1 and 0.9 or does it happen simultaneously at 1.0? This could be tested by giving A a 1 CE headstart and hitting Provoke nearly simultaneously. Then slowly ramp up the CE headstart A gets until he keeps hate from his Provoke, even though B is after.

Also, do all mobs check their hate lists at the exact same time? Get two groups of 3 players together and do the /kneel test on 2 different mobs, see if they turn towards the kneeler in sync.

Basically, I'm thinking identifying the tick and the smallest chunk of VE the game works with would give us a multiplier for every action which we could then use to round to the nearest possible number based on observation.

(I would do all this myself, but I'm having trouble updating FFXI, so I live vicariously through others ^^)

[Kaeko]

Kaeko, when you did the test where Ashira /kneeled and the mob would turn towards her and then back... did you ever note the frequency of the turns? This might be an important clue to the minimal unit of VE. The enmity tick (however long it is, 1, 1.2, 2,4 seconds perhaps) is imo fundamental to understanding exactly how VE decays. -60 VE per second is a great start, but is it continuous or discrete? If it is continuous that would imply there is a counter on each character ticking down 1 VE at a time. Personally I think it is discrete. Picture the following:

Each player has his enmity counter. At every "enmity tick" the mob simply goes to each player in his list and "eats" an X amount of VE from that list. At the moment it manifests itself as -60 per second, but if the "enmity tick" is perhaps 1.2 seconds that would mean the mob eats a chunk of 72 VE per tick. Also, viewing decay as discrete would explain how identical actions beginning at different times between ticks would still have the identical VE and wear off at the same tick.

For example, say the tick is 1 second. Mob checks his list at 0.0 and sees equal enmity on 2 players. Player A performs an action at 0.1, player B performs an action at 0.9 What exactly happens here? Should B get immediate hate? Has A's VE decayed between 0.1 and 0.9 or does it happen simultaneously at 1.0? This could be tested by giving A a 1 CE headstart and hitting Provoke nearly simultaneously. Then slowly ramp up the CE headstart A gets until he keeps hate from his Provoke, even though B is after.

Also, do all mobs check their hate lists at the exact same time? Get two groups of 3 players together and do the /kneel test on 2 different mobs, see if they turn towards the kneeler in sync.

Basically, I'm thinking identifying the tick and the smallest chunk of VE the game works with would give us a multiplier for every action which we could then use to round to the nearest possible number based on observation.

(I would do all this myself, but I'm having trouble updating FFXI, so I live vicariously through others ^^)

I think that these "timing" related things are important for VE because of that decay rate which is time dependent; however, I don't think I'm knowledgable enough in machine code clock cycle stuff to really mess with that on any deep level. I'm sure a lot of others on the forum can try to set up a controlled test for that and would do a much better job of it.

I think for now, I plan on just leaving "margin of error" on the things that involve VE. It's really hard to tell the exact VE of a spell, how often something checks which person has highest enmity, etc. experimentally. Anything that involves a stop watch in the test is going to lack really good precision, but that's as good as we can get it without knowing the details of how clock cycles work in the game.

For the healing thing, I can tell you it doesn't just check every "tick" of HP/MP healed - it checks what seems like every few seconds maybe (1 vana minute?). I'm not too sure but it doesn't seem to follow normal properties. For instance, we had 1 player at +1 CE over the other. The player behind then /healed and would take hate momentarily every couple of seconds, which is to be expected; but, the hate was shed incredibly fast and the mob returned to the orginal target extremely quickly. I don't think the VE decay rate applies to this - /healing will easily take hate even +100 TE over the other player - by decay rate, it should take at least 1-2 seconds to come back to the first target, but it does not.

[Liselle]

Sounds like a job for FRAPs. The ability to look at the actual frames of a video will help precision, and may also give you a clue as to how often hate "ticks" (in the case of /healing, for example, may help solve that mystery).

It's still not going to be as good as the CE testing, because unavoidable lag is going to subtly alter things. However since reaction time does not enter into it at all, and you can see the EXACT moment that a mob starts to turn, it should be more precise than a stopwatch.

Just something to think about if you are looking for a way to come at this VE and hate decay problem from another angle.

[Ashira]

Sounds like a job for FRAPs. The ability to look at the actual frames of a video will help precision, and may also give you a clue as to how often hate "ticks" (in the case of /healing, for example, may help solve that mystery).

It's still not going to be as good as the CE testing, because unavoidable lag is going to subtly alter things. However since reaction time does not enter into it at all, and you can see the EXACT moment that a mob starts to turn, it should be more precise than a stopwatch.

Just something to think about if you are looking for a way to come at this VE and hate decay problem from another angle.

Hmm, not a bad idea; the biggest problem here would be our machines.. Kaeko has horrible lag and mine doesn't deal with FRAPS well - we had issues doing some of the tests at all because he wouldn't always see the mob turn when I did, and there was once I didn't see a turn he swears he saw. Not sure of anyone who has a good machine who'd be willing to do nothing but stand around and hit "record" and "stop" for a couple hours with us.

Maybe we'll get lucky though and someone here can help out in this manner as we progress on all these tests, and see if we can't narrow it down a bit more.

@Valyana: did you get any better results with your Sirvente tests?

[VZX]

Computer is a state machine >.> Whatever you see as continuous is actually discrete.

and yeah... FRAPS is a good start to analyze it more detail (like what I did on ranged delay test).
Just give mark to the first frame mob trying to turn around between 1 player to another and mark it again on frame where mob is trying to turn back.

[Valyana]

Some more stopwatch testing:

Dispel, Sleep, Sleep II seem to have the same VE as CE.
Warcry gives 1 CE and ~300 VE per player affected.
Barfira gives 1 CE and ~80 VE per player affected.
Shield Bash for 12 damage gave 1 CE and ~900 VE. I was surprised that it was only 1 CE since it did a large chunk of the Tiny Mandragora's HP in damage. Maybe Shield Bash damage is hateless?

Pulling with a spell gives +200 CE:
BRD pulls with Silence
RDM Stun + 21 cures to pull hate -> 201 CE

It doesn't seem to give a lasting +Enmity bonus, as after a Dispel from each of us we still had balanced CE. It does seem to give a significant amount of VE as well, but I'm not sure how much.

I'm not sure how this fits in with /healing mages stealing hate from a pull with missed ranged attack. Perhaps a missed /ra doesn't get the +200 CE bonus.

Pulling with Finale + 20 cures = Dispel, so Finale gives 100 CE.
Stopwatched at ~240 VE.

I'll check other BRD songs later.

I suspect that Finale and NIN debuffs CE and VE really add up to the same number as Paralyze et al VE, rather than being ~20 different.

I also repeated the L3 Sirvente test with the same result as before.
BTW, Zozma and Dragonade are helping me with my testing.

Edit: only tested Warcry and Barfira up to 3 players, but I assume they scale up to a full 6-person party.

[Hanyoko]

Considering that players can have multiple 'tick' timers running on them (hmp/hhp, regen, refresh) its probabaly realistic to assume that the hate decay timer is actually on the mob.

If the mob is the entity that tracks the enmity of people on its hate list this would support how it can remember you if you zone/log.

On the other hand if the turns relate to when the player hp/mp+ tick happens then its the player entity updating the hate list.

Or it could be something more complicated and having VE on the player and CE on the mob and the one 'polls' the other at random times and updates their listing on potentially 2 unsync methods.

[Ferien]

Shield Bash for 12 damage gave 1 CE and ~900 VE. I was surprised that it was only 1 CE since it did a large chunk of the Tiny Mandragora's HP in damage. Maybe Shield Bash damage is hateless?

That's interesting. Should probably test with an Aegis as well, possibly the enhancement changes the way its hate functions? It would seem counterintuitive to think that the 230 damage or so is completely hateless.

[Devinry]

Any indication on whether or not Dispel needs to "land" (i.e. actually dispel something) in order to generate the CE? I know Stun doesn't have to land to earn hate, and I noticed that your testing with Aspir was for 0 MP. Just curious if the same applies for Dispel.

Apologies if this has been answered already, and Kaeko/Ashira (/everyone else), you rule for doing this.

[Valyana]

Any indication on whether or not Dispel needs to "land" (i.e. actually dispel something) in order to generate the CE?

It doesn't.

[Khamsin]

Any indication on whether or not Dispel needs to "land" (i.e. actually dispel something) in order to generate the CE?

It doesn't.

I don't think there's any action you can perform that causes no hate if it fails to do anything. Even curing for 0 damage or a miss against an enemy causes hate.

Except Gauge, I suppose, but that doesn't really count.