Poison Creeper

[electricblue]

this is to be a part of my summoner guide, but I think I should publish this before the launch of the guide since people keep asking questions about the creeper.

here is the url:

www.geocities.com/welett/creeper.htm

comments, inputs?

and if this suits the workshop rather than here, sorry than.

[Brista]

The short answer is to point to the Arreat Summit: 84-86 poison damage over 4 seconds at skill level 20

(The Arreat Summit is pretty accurate these days)

Chippydip's calculator goes past level 20, I've lost track of where that's hosted now

The longer answer involves getting the damage properties from the .txt files and calculating it from there. Poison damage is handled in small fractions of whole damage points so any figure like 84-86 is actually an approximation of what's really happening

Edit: ah sorry I see from your write-up you have this data already

So what you're really asking is when poison creeper acts in odd ways like the sweet spots and apparent stacking, what is in fact occurring mathematically?

That's a much more interesting question. It appears to have an effectiveness out of proportion to its low listed damage values - I've had a 25ish creeper obliterate stuff at the end of normal in large games, far out-damaging twinked cookiecutter builds

[electricblue]

Um, in fact I am just sharing some new finding on Poison Creeper and asking for comments or more inputs.

[Deadpan]

I had a lot of fun aggressively casting Poison Creeper with my 1.09 publican HerbalRemedy, licenced to serve wines and spirits - variant named by Brista, as it happens ;).

My vine almost never died from (un)natural causes and I was still chalking up vine kills in Act 5 NM. Totally unscientific, I know, but I kept spamming in the belief that it kept the lag to a minimum. I was getting quite a decent duration on my poison mats and each time I recast I was creating some more so stacking clearly worked. The build was obviously too uber for Blizzard - I also mastered the art of stacking spirit auras - so they made sure that the patch nerfed it big time. They fixed the lag though.

[Baajikiil]

Um, in fact I am just sharing some new finding on Poison Creeper and asking for comments or more inputs.

You may want to mention the fact the the Creeper(along with the other vines) is only vulnerable to melee attacks. All ranged attacks(such as Diablo's LBOD, and nova) pass over without damaging it. This makes it a decent distraction for ranged monsters, and allows it to actually kill act bosses(given enough time).

A good tip to aim the vine at a monster is to recast it right on top of something. If cast next to a monster, it will immediatley attack, leaving the poison mat right where you want it(like a doorway in act2, or on top of a leader).

As for it's unusual damage, my experience leads me to believe that poison creeper's damage is completely mistated. My best guess is that it does the listed damage *every second*, not over the course of 4 seconds. It seems like it was doing 4 times the listed value in the same amount of time when I used it. this needs to be tested under controlled circumstances though...

(I played a druid who used poison creeper exclusively for offense all through normal. Unfortunately, he died on an unlucky Baal run. He was able to solo diablo, baal and the ancients using only poison creeper.)

[electricblue]

poison creeper is useful for mass targets, not on a single target, so, no killing of act bosses.

also it does its listed damage to a single opponent. things change against multi opponents.


I dont think there is an upper bound for the damage of poison creeper, but in legit possible, it seems to be around 10k/sec

[Brista]

10k/sec for a skill that lists 20/sec at skill level 20 is a pretty large claim

Could you give details of your testing please?

[Brista]

so, no killing of act bosses.

Diablo in Normal has 13 818 life

A poison creeper doing approx 20/sec will get through that in just over 10 minutes, so it's certainly possible to do it even without skills gear

[adeyke]

With perfect +skills gear and Battle Command, it looks like you could get about slvl 52, or slvl 54 if skill shrined. Those would give about 79 or 82 damage per second, respectively.

[electricblue]

still very inefficient for a single opponent.

[electricblue]

10k/sec for a skill that lists 20/sec at skill level 20 is a pretty large claim

Could you give details of your testing please?

infect a single oopponent 11 times (or as many as the number of mats you want to effect) with a lvl 50 (or whatever lvl) creeper, than move into the mat a second opponent, note the damage.

[Brista]

Now that is VERY interesting

Generally in this game poison from different sources doesn't stack. If a Venom assassin hits a monster poisoned by a team-mate then the strongest poison effect will take place and the lesser will be cancelled. (Highest damage per second)

What you're saying (and I think you're possibly right) is that the poison mats DO stack

Can I ask for a bit more about the tactics needed to bring this off? Do you think that basically just casting a creeper at a target and then recasting when the mat appears is the most effective way?

Have you tested whether the creepers stack with other Poison effects such as Rabies?

[Baajikiil]

infect a single oopponent 11 times (or as many as the number of mats you want to effect) with a lvl 50 (or whatever lvl) creeper, than move into the mat a second opponent, note the damage.

I hope someone can do this testing in other than pvp. Pvp adds issues that make it hard to get to the true values. I didn't notice anything crazy like you have when playing with poison creeper in pvm, but it's possible I simply didn't recognise the effect when it was occuring.

BTW, did poison resistance work properly against this when you tested in pvp? How about -% poison duration?

After stacking the mats "11 times" how long does the poison last? 4 seconds, or was it more?

Perhaps it's time I remade Poison_Creeper for the new ladder.

[electricblue]

I am preparing a summoner guide. There will come more and tactics.

BTW The test was done only to determine the damages.

[Crystalion]

Now that is VERY interesting

Generally in this game poison from different sources doesn't stack. If a Venom assassin hits a monster poisoned by a team-mate then the strongest poison effect will take place and the lesser will be cancelled. (Highest damage per second)

What you're saying (and I think you're possibly right) is that the poison mats DO stack
...
Have you tested whether the creepers stack with other Poison effects such as Rabies?

Ah, but technically poison *does* stack, on the slice of poison delivered with the initial attack. You would normally do just fine ignoring this effect because it is unusual for large numbers of attacks with poison to connect in a small number of frames. But if I made a mod with an aura, like Rabies, delivering (aka "renewing") poison every frame or something like a Blade Sentinel minion without a nextDelay then clearly I would see the significant possibility of "stacking" poison attacks.

I've often noted that the Poison Resist Shrine is my favorite, because, by forcing poison durations to nada, it *almost* makes you immune to poison. A reason it does not, is that initial slice of poison (when you are first "hit"). If we made stacking+every_frame poison applicators then the Poison Resist Shrine would be effectively bypassed.

Thus my reaction, upon seeing this thread when first posted, was that it seems possible the "mat" is an entity that frequently reapplies poison to the victim. In previous Creeper strat guides I'd seen the suggestion that monsters needed to be in motion to trigger the mat effectively... I strongly suspect the circumstances (aka multiple monsters, or monsters moving) interact in some way with the application process to result in "stacked" repetitive "initial hit" poison application.

To test this I would suggest lowering the duration from 4 seconds to one frame but upping the damage 100x. This, in some (mathematical) sense, should "preserve" the net damage, but I bet the actual damage patterns and visual (green victim) would be more readily apparent (as you fool around with monster movement, multiple monsters, or whatever else might be a factor).

edit: the Creeper skill creates a minion (the plague poppy vine) who fires missiles (which fire missiles... aka the mat, of which the "plague vines trail" appear to do the dirty work). These things appear to have quite a range (size "3") and at high slvl there appear to be a lot of them per mat (I'm unclear on the number as I see two different places apparently defining it). At the very least these dozen (more?) dirty work vines are possibly each attacking the same target (the duration poison would not stack, but being hit 12 times for 1%, i.e. one frames worth of the 4 sec. x 25 fps poison nut, would be measurable I'm sure).

My recollection (deserves testing) is that if a poison cloud (and probably mat) hits you and you don't move, the duration will end after X (i.e. 4 for mat) seconds. That is to say, there appears to be some game mechanism recalling if you've been hit and not hitting you again. My naive guess would be that missiles flagged "LastCollide" cause a simple hack whereby the missile records the entity ID and x,y of a collision, and grants "immunity" if it's next (and subsequent ad infinitum) hit attempts are to the (exact) same. This would probably be a single bucket, so a missile that does not self-destruct (most of them CollideKill, ergo collide and cease to exist) faced with persistant multiple targets would be able to repetively hit (either alternating frames or doing both each frame depending on the implementation logic).

My present inclination for a "good" test to probe these hypotheses would be to use a stationary PvP victim, a zon with her side-by-side decoy (an no decoy, of course, for comparision), and throw Poison Creepers of various slvls at her feet (more sub-vine missiles per slvl, up to some max, you see). Given PvP reduction, bumping the skills.txt poison str by, say, 10x might be expedient.

p.s. other "oddities" recently on my mind... a posted claim (in passing) that Blessed Hammer benefits from Might and Fanaticism *if* the user is being affected by Concentration; my claims, that I've never seen follow up on: that straight elemental absorb seems to be granted life both before *and* after damage dealt; that lifetap with thrown weapons seems to grant absurd multiples or overage of stealing.

[Crystalion]

My present inclination for a "good" test to probe these hypotheses would be to use a stationary PvP victim, a zon with her side-by-side decoy (an no decoy, of course, for comparision), and throw Poison Creepers of various slvls at her feet (more sub-vine missiles per slvl, up to some max, you see). Given PvP reduction, bumping the skills.txt poison str by, say, 10x might be expedient.

v1.10 release, modded skills.txt Poison Creeper skill to 250 for both Emin and Emax and 0 for all the elvl increments (in other words, I don't up the poison via slvl, but I do affect the #sub-vines). Changed E duration to 25 (one second)

-act 5 pvp test with

victim: vit pumped clvl 33 zon with slvl 10 Decoy
attacker: druid with slvl 1 (modded) poison creeper

procedural note: since Unsummon on vines doesn't seem to work, I put a point in the next vine (the life reclaimer) so I could "unsummon" that way.

I tested by summoning a Creeper at the feet of the target and waiting for the mat/attack, then immediately unsummoning (as noted above).

For the non-Decoy test this resulted, as expected, in a single ~second green zon who took very little damage (note the mat lasts for several seconds, but clearly did not re-poison the stationary zon).

For the Decoy test this resulted, not surprisingly, in effects contrary to "normal" understanding but similar to expected from the hypotheses posted above... the zon and decoy were green for the whole duration of the mat's existance, they grunted a lot, and the zon took huge amounts of damage (greatly over 10x the prior case).

Note that this is, I believe, a "3 mat" case (slvl 1 Creeper) with no stacking of mats involved.

If, when I test the higher slvl (more sub-vines) or stacking (casting multiple creepers) I see anything other than expected (i.e. correspondingly more damage with Decoy, but modest additional damage sans Decoy) I'll report back.

edit: well, at slvl 20 the duration of the mat is quite long, so the Decoy is oblitered in a few seconds and then the zon stops taking damage. To rectify this impediment to testing I brought in a 2nd zon, also clvl 33 with 100 vit (iirc 354 HPs, both zons naked) and stood them by each other (Druid hostile to both). No grunting (must have been the decoy) but both zons then die in only a few seconds (much shorter than the duration of the mat).

Interestingly this suggests yet another possible entry in my "faux PvP trivia" file: using a poison immune minion (recast as needed) next to a big boss (act boss, e.g.) you'd have two targets for the mats (aka sub-vines) and would therefore trigger the massive poison damage anomaly (the enabling minion would not actually care about the mat, being poison immune, but I suspect the mat would still attack them and thus be enabled to hyper-attack the boss).

Coincidentally I'd been considering building (another) rabies druid for fun (seque to fury, in order to be viable in Hell)... my thinking had been that Poison Creeper was ultimately a waste (i.e. post normal) other than as synergy, but now I'm thinking that Rabies is somewhat of a waste, given spammed maxed mats. Note that the number of sub-vines caps and the ramp up of creeper damage is slow, so once one gets a good duration, it seems (at first intuition, ymmv) far more important to control mat placement and "stacking" than to burn effort getting super high slvls for the skill.

If my hypotheses are indeed the underlying explanation for this observed phenomenon, then the "size 3" is a key constraint, essentially requiring monster crowd (i.e. pair) placement within that radius for each sub-vine you want to (repetively) hit.

To position enemies exactly in "the zone of death" (i.e. bunched very close, pair-wise, on existing spammed mats) seems to me a task that would bias Druid behavior significantly (e.g. Grizzley = bad, because of knockback; Druid = sole tank perhaps good, due to melee-type enemies clustering around him in the sweet spot). If even seems reasonable to me that a druid might forego wereform just to legally be able to cast charges of Attract (a great cluster tool).

Naturally, in party play, this cluster desire is not unknown (CE, Meteor, etc. etc.), but largely I don't think people think much about it (as clusters happen naturally and frequently for most party M.O.s that I've seen in any event).

Since 20 points in Poison Creeper isn't a huge investment, perhaps a Wind Druid build might consider it as a second form of damage.

Of course, given that slvl 1 charges recharge fees are low, I suppose this means that the Carrion Wind unique ring is situationally quite useful (the problem being, like items with charges of CE, that the number of charges is inconveniently low, and few people enjoy TPing to town often).

I suppose I should make a test barb with Carrion Wind and have him Taunt and play tease to check if a monster walking through the vines (i.e. alone, no pair in size 3 range) moves fast enough to receive a lot of damage (as per hypothesis--or if this part of my musings is off-base). I'm too lazy to do that atm, however.

edit2: making the Decoy poison immune made no difference to the effect (other than the decoy never taking damage, of course); I played a bit with spacing between the zon and decoy and the "size 3" theory certainly looks like a good fit; the creeper appears to be willing to lay a new mat as frequently as every couple of seconds (I see a 15 frame factor in the mpq, so perhaps the AI takes a 15/25ths second rest after an attack)... with the duration of the mat up (slvl) I could clearly see these had "stacked" on me (once I cast a decoy to activate the evil)--I would think the creeper is frail enough and it's AI fickle and slow enough that spamming new creepers at the feet of the target to stack mats upon is probably a far better idea.

Naturally critters like ghosts and cows that bunch up in herds would be very good candidate targets for this effect.

[Brista]

Splendid research!

Thank you both for your efforts

The main druid cluster technique I can think of aside from personally tanking is lobbing minions into packs. Grizzly is decent for this as it can take a while to be dealt with, Oak Sage can work too. Alternatively spirit wolves - with +4 skills for the full five you could keep spamming them into a pack of monsters

I've only taken Poison Creeper as far as late Nightmare, can anyone attest to using it successfully in Hell?

btw: could the druid's own minions be considered as extra monsters by the program for the purposes of calculating the traffic over the vines?

[Crystalion]

I've only taken Poison Creeper as far as late Nightmare, can anyone attest to using it successfully in Hell?

btw: could the druid's own minions be considered as extra monsters by the program for the purposes of calculating the traffic over the vines?

Well, if I understand what I'm seeing here, then the basic damage potential is related to the #sub-vines you can get "adjacent" to two targets (enemy! so your own minions don't help, as the vines do not target friendlies and so will not "forget" an enemy target in order to zap them next frame)... this is a factor of slvl (until the cap max. #sub-vines) and of spamming (how many mats you can lay down on top of each other, either by creeper recast, or single creeper re-attack).

This ought to mean that less than 100x the normal stated damage is practical. The normal damage of creeper is so so low by Hell standards that I hesitate to endorse even 100x!! This is completely aside from the issues, ala Blessed Hammer, of positioning this "great attack" so that it actually connects.

I can not think of a related line of research, unless, perhaps, spamming Plague Javalins might experience a similar effect (I've not looked this up for plausibility nor tested it). Most AoE attacks are meant to hit every frame or have a NextDelay. That a poison based AoE would have this oddity is, of course, likely to be overlooked, due to "poison doesn't stack" being approximately correct. Certainly someone could examine every missile (missiles.txt) to see if there is another such quirk (perhaps MvP).

I do think it is clear that this oddity is powerful enough, when understood, to explain the rapid leveling in normal of obvious pre-Rabid builds, dispite the awful damage numbers for the Creeper. Since quickly leveling to 20 or 40 has some utility (non-Classic rune rushing) this is of some interest, being, like the FireBlast/WoF race build, essentially independent of equipment. (I do not mean to shaft other race to 20/40 builds here by not mentioning them, it is just that I've done FB/WoF a number of times myself and find it really easy the whole way).

One test I did not do, that should be done, is to verify that the mats succeed in delivering their damage even though they *never* overwrite a bigger rate of poison on the target (i.e. a confirmation of my theory that the "hit slice" of the mat poison is the key here, and not the duration poison). Should this prove to be true then the mat technique (pairs of monsters) doesn't offer or suffer interference with other poison techniques (Rabies or whatever) making it valuable supplementally, even if the verdict is that it is too weak for Hell.

[electricblue]

highest poison rate is applied for its duration if another source exists.

thnx crystallion for the research. One important point that should be searched (which I can not do) is this if you can:

what happens when 2 or more guys enter into the layered mats exactly at the same time.

in PvM, I have witnessed some extreme amount of damage maybe caused by this, but not sure.

more on creeper will come with the guide, but here are some numbers.

hell pits take about 8 mins
hell cows take about 15-20 mins
only using the creeper and myself as the meat shield

[Crystalion]

highest poison rate is applied for its duration if another source exists.

thnx crystallion for the research. One important point that should be searched (which I can not do) is this if you can:

what happens when 2 or more guys enter into the layered mats exactly at the same time.

Given what you've said about Hell play, it seems pretty clear to me that PC is an uber skill for norm/nm, when "properly exploited".

If dumping all the charges of Carrion Wind at once, vs. a nasty pack, isn't too spendy to repair, then that's certainly an interesting toy for high level players (probably more effective than the "Returned" effect).

But if you can't get enemies close together, it just isn't going to fly. In that sense the skill is probably weaker than slvl 1 (i.e. low radius) CE or Static Field in the "balance" department.

While I note many people looking to use CTC SF equipment, I've not heard of anyone trying to make good use of runeword Black (for CE), for instance. So I suspect your interesting discovery won't get used very much. OTOH a Fury werewolf build certainly can spare 19 points if it helps the kill rate in general (ala PMH) and vs. physically resistant or immunes in particular.

...

I was in "research"/mumble mode, so perhaps I should be a little clearer in what I believe to be true (as opposed to mixing my reports of observations and hypotheses chronologically)...

Poison is applied in an odd manner.

First off, the LCS (Lying Character Screen™) is misleading, because poison really is always in a per frame scale.

Although I don't recall that Jarulf made it clear one way or the other, I believe that "per frame" poison value (after resist) is summed into the total damage calc for an attack. (Since I'm not going to dig up his most recent relevant post nor find the code in question, you'll have to suffer along with my educated guesses...)

*Later* in special processing, that value will be compared to the target's existing I-am-poisoned-this-much-per-frame value, and overwrite it (and the frames length, appropo) if the new stuff is "better" (stronger per frame). In the case that they are exactly equal I do not know if the new length always overwrites, never overwrites, or overwrites only when the length is longer than the existing remaining length (I suspect that it is this latter). I also expect that either at this point, or via aura-like "pulse" the length will be zero'ed if the target is under the influence of a Poison Resist shrine.

The point of all this is that the initial contribution of poison to the total attack damage is usually pretty small compared to this latter effect of ongoing poison. This leads people to the simplification that "poison doesn't stack" (because the latter effect does not--mathematically it would be quite easy for it to stack "properly" in some sense, but earlier versions of D2 goofed up here and Blizzard simply stopped poison from stacking instead of getting it right).

However, if we had a zero duration poison attack to toy with (i.e. like Holy Freeze adds zero duration chill with the added cold damage) then there would be no question that poison does indeed stack (i.e. the initial single frame damage).

Now things get interesting. Because, basically, you've found a case where a poison attack will be repeated every frame, afaik.

More than that, you've found a case where a "stackable" (aka spammable) *cluster* of poison attacks will be repeated every frame.

Now these are all of the same magnitude, so if it wasn't for the tiny single frame slice of the poison damage that is applied immediately as "part" of the total damage of an attack, none of them would do anything effective, other than to keep the poison duration up (which, e.g. I believe the Rabies aura does, but, iirc, the aura applies a length which is decrementing, so it doesn't seem like it--you can discover it, however, by "curing" yourself with an antidote potion and observing your re-poisoning).

The reason, afaik, for your unusual case, is that preventing some stationary missiles from hitting the same target every frame is done by an imperfect hack (kissing cousin to the NextDelay hack). This simple and efficient hack appears to be simply storing in the missile the information of what its last collision was (possibly where and with what) so that future answers to the question "what have I collided with now?" that are the same can be discarded (in order to avoid repeatedly hitting a stationary (?) target).

There are flags for the missiles that either use this hack, or the NextDelay hack, or some other collision decision process depending on what the designer intended the effect of the missile to be. But, for example, although firewalls, shockwebs and poison mats all lay in one place on the ground to zap those upon them, the designer had a different goal (and flags therefore) in mind for each of them vis a vis their repeat hit behavior.

If I changed the damage type of a Firewall to poison, you might guess, since everyone knows "poison doesn't stack" that standing in it vs. passing through (and being poisoned by it) would have no difference in net damage. I believe such a guess would be wrong. (The reason, again, is that one frame of damage is simple applied with the attack, and a firewall attacks every frame).

So these missiles that don't move that make up a poison mat don't do much more damage than expected to a single stationary target, because the hack flagged causes them to remember to not hit that target again (jRND thot: I should have tried tiny teleports).

However these particular missiles have a very large collision radius (size 3). This means a given missile can readily hit a number of nearby targets (given the typical size of targets). When this situation occurs, the hack fails to perform as intended (i.e. implementaion obviously falls short of design intent).

So the answer to your question:

what happens when 2 or more guys enter into the layered mats exactly at the same time.

is that any time a missile finds itself in range of multiple valid targets (i.e. enemy) the hack will fail. Since I don't know the exact coding details, I'm guessing that this results in either 12.5 or 25 attacks per second per missile per target (assuming two targets in range).

Each Mat has a lot of missiles (apparently 3 at slvl 1 and another per slvl up to some max #). Knowing of the situation you're arranging to spam mats. Conceivably many dozens of missiles could be in range of a particular target (and another target).

Maybe at one time they ran into this "bug" with Shock Web and that's why they slapped such a ghastly NextDelay factor on it.

Since Blizzard has a history of balancing games "bugs" included (the ARx4 thing being a famous example) I'm sure the pathetic ramp up of damage for Poison Creeper vis a vis nm/hell is as it is because vs. crowds it is still somewhat effective (though I doubt Blizzard understood why and so I doubt they realized it was much more effective spammed).