1.10 Monster mlvl and TC upgrades

[Ruvanal]

I finished testing some changes in the way the game is handling the monster upgrade of the TC that they use for item drops. The findings are here.
http://phrozenkeep.it-point.com/forum/view...pic.php?t=15949

Also from another thread at Phrozen Keep,
Ruvanal:

 
more information on a couple things relating to the monsters.
Monstats.txt
noRatio: Do not use the ratio factors from monlvl.txt for the AC, TH, HP, DM and XP factors. Just use the straight values from this table instead.

Extra material: The noRatio flag is also used as a flag for these two aspects of the game.
1. If noRatio=0 and the difficulty is nightmare or hell, then the mlvl of the base monster when spawned is set to the higher of the monstats.txt level for the monster or the level of the area from levels.txt. If the difficulty is normal or noRatio=1, then the unmodified level from monstats.txt is used.

2. If noRatio=0 and the difficulty is nightmare or hell, then the assigned TC from monstats.txt will be modified if the TC in TreasureClassEx.txt is part of a group and the level of the assigned TC is less than the mlvl of the monster. The new TC that is assigned will be from the same group and and the highest level that is less than or equal to the new mlvl of the monster.


Example: someone puts quillrat1 (Level(H)=67, TreasureClass1(H)=Quill 1 (H) )from monstats.txt to appear in the hell difficulty Arreat Summit (id=120, MonLvl3Ex=87). The actual level of a non-champion, non-minion, non-boss Quill Rat there would end up being mlvl=87 and would drop the Quill 5 (H) TC when killed.

This kind of change in the way that the monsters are being handled is part of the reason that Arreat Summit took down their listing of individual monster stats. There would be too many listing required to reflect that some monster could be varing in suttle ways from area to are that they end up appearing in. This change also will cause an accuracy problem with the information at this site.
www.d2data.net

Something to think about when going into some of the areas or when you see a drop that looks out of place.

edit:fixed link

[Crystalion]

Example: someone puts quillrat1 (Level(H)=67, TreasureClass1(H)=Quill 1 (H) )from monstats.txt to appear in the hell difficulty Arreat Summit (id=120, MonLvl3Ex=87). The actual level of a non-champion, non-minion, non-boss Quill Rat there would end up being mlvl=87 and would drop the Quill 5 (H) TC when killed.

Very interesting. That makes it somewhat hard for a neophyte like myself to tell what act 5 monsters might be ideal for MFing certain sets and uniques. Hopefully someone will put together a better guide than just looking at the misleading data at d2data (which still, btw, has omissions in some sections, like affixes, for v1.10).

So, I if understand this correctly, Revivemancers potentially benefit from reviving preferred monster types albeit of a higher level, because they are "guest" monsters? Or do Necro's revives always get clvl for mlvl or some other formula not related to original mlvl?

I hadn't considered this issue before in rushing, but clearly a reviver rushee can fight more easily above his own level if his revives are comensurate with what they are fighting. Or better still, he's rushed to act 5, darts out and grabs some freebie corpses, then waypoints back to a level appropriate act to fight, potentially starting a new game any time his revives expire and he needs more.

(general note: the pindle corpses are on timer and will die at timer trigger if they are revived--however you should still be able to poison explosion them. But Prowling dead are 90% poison resistant so it won't help much. Corpse Explosion works, but timing prevents it from helping as the newly risen PD are invulnerable for a time and the timers aren't staggered enough for there still to be corpses left when they become vulnerable. Barbs dashing in and horking these corpses for items is pretty funny--no idea if any of this works on realm; all tested only SP).

edit: upon reflection, I imagine the mlvl of a revive must be set = clvl. This would still, however, beg the question of whether guest monster scaling in v1.10 might make Necro revives (some monsters at least) studlier than pre v1.10?

[Crystalion]

The findings are here.
http://phrozenkeep.it-point.com/forum/vi...hp?t=15949

The link isn't working for me. I did read some of your recent posts at PK though. This bit puzzles me...

nightmare Cold Plains monster that were dropping qlvl80+ items while I was testing the chests. The testing that did on that showed the following:
In nightmare/hell difficulties if the base monster type has monstat.noRatio=0, then the game will set the mlvl to the higher of the area level from levels.txt or the listed mlvl from monstats.txt.

because I don't think nm Cold Plains would have an area level of 80+ so I'm assuming a guest monster did--but how does a *nm* guest monster have level 80+?! This all makes perfect sense to me if you meant you were testing hell, not nm.

edit: while I'm at it, I have a question implied in another post of mine, based on something you wrote at PK:

With cube codes and functions that is correct. But there is a loophole to that condition that Isolde introduced into v1.10 that will allow ilvls up to 127 for items; monster drops.

Since, iirc, I had a level 99 char buy an item in a shop with 71 charges of level 11 sacrifice. According to theory, at ilvl 99 it should be 67 charges of level 10, right? So my presumption was that either rounding (up) or ilvl > 99 was responsible. In a nutshell, the question would be: does v1.10 allow shopping of items with ilvl > 99?

[Ruvanal]

I fixed the link.

The nighmare thing came as I was testing the chest drops for TC upgrades. I had set the level of the Cold Plains to a much higher level than it should have been in nightmare to see if the TC that chests were using was getting modified. The items dropped from annoying monsters that were in my way, so I used ATMA to examine their ilvl. When it was too high for what would have been expected there in the first place I started looking for the cause. The answer of which is in the content of the first post here.

For the ilvl matter, Isolde had stated in thread at Phrozen Keep that he had only removed one of the ilvl caps in the code. The only place that would fit from the current observations would be the item level of the monster drops, as the other sources still do not appear to allow an ilvl of greater than 99. With the store ilvl=clvl+5 and the character you are using the ilvl would have been 104 if it was allowed. If you try an ilvl=104 in the formulas that Hammerman listed you still would not be able to get a slvl of 11 out of it. I suspect that Hammerman made an error at one point and that there is +1 factor that he missed including at some point in the formulas he was listing. It would not be the first time that I found that type of error in one of his listings (it is fairly easy to make some errors in translating the assembler code to a 'normal type' of psuedo code). There also may have been a subtle change in the formula from 1.09 to 1.10, but I doubt that.

Edit: I checked the vendor item generation and it will allow greater than ilvl=99 now. But that still will not allow an ilvl=104 of generating the slvl of 11 that you observered based on Hammerman's formula. The game is using integer divisions which will only truncate the fractions, not round them. It appears that there is just a missing "+1" factor in what he listed.

[Crystalion]

Edit: I checked the vendor item generation and it will allow greater than ilvl=99  now.  But that still will not allow an ilvl=104 of generating the slvl of 11 that you observered based on Hammerman's formula.  The game is using integer divisions which will only truncate the fractions, not round them.  It appears that there is just a missing "+1" factor in what he listed.

I don't see that there are really any possible effects (better affixes) possible because of this, but I'm not really up on all the alvl possible or not stuff.

Looking for an alternate explanation to "rounding error" in the formula suggests the following...
Hammerman:

If "max" is a negative value, it does:

slvl = (ilvl - uselvl) / neg(99 - uselvl / max)

99 - uselevel is always at least 1. it also makes sure that neg(99...) is at least 1 before the divison.
uselvl is the highest of reqlvl from skills.txt and level from affix.txt.

Since the skill level req is 1, if the level used from affix.txt were the *class level* and not the general level req (1 vs. 7) then all the observed data would match the formula...

gamble ilvl cap at 99: (99-1) / -((99-1)/-10) or 98/9 or 10 when everything gets truncated at each step.
shop ilvl at 104: (104-1) / -((99-1)/-10) or 103/9 or 11 ditto.
and, of course, using a uselvl of 7 doesn't exhibit this boundry crossing. Just a thought.

edit: indeed, using a -act 2 level 16 assassin to shop drognan shows level 2 sacrifice charges, ala...
(21-1) / 9 or 2, rather than (21-7) / 9 which ought to truncate to 1.

What you think?

...

Purely for historical/testing interest: before I saw your edit I gathered additional information that shop bought ilvl exceeded 99, as follows...

I modded of sacrifice to have high magic_suffix.txt (column:) frequency (so I'd see a lot of them and not spend forever testing--I tried without at first and was going crazy). My level 99 character then observed scepter type weapons from all nightmare shops with 71 charges of level 11 sacrifice.

I then modded gamble.txt to have only scp (making it easy to buy lots of scepters, which don't normally gamble). I did this because I suspect Firestorm, which can gamble at the expected 67/10, would show the same 71/11 if I were to do that test, since I suspect "10" might be sufficient to see rounding effects (I've never observed another charged item coming out contrary to prediction by Hammerman's formula).

I looked at lots of gambled scepters with of sacrifice, and they all had 67/10. My sample was big enough that I should have seen at least one 71/11 if the effect were possible at ilvl 99.