A week ago I posted a forum thread that served as a survey of fan opinions on item drops in Diablo 3. It got some immediate attention, assisted by Blizzard’s promotion of it, and over the weekend I read over the dozens of replies and compiled them into this article, which summarizes the results and general themes of fan comment and complaint. It also breaks news, at least news to me, in terms of how the game selects specific legendaries to drop, and why some of the best ones are so spectacularly hard to find.
There were eight questions in the survey, asking how players felt about the quality and quantity of Diablo 3 legendary items, what Magic Find levels were needed for success, if set items were too uncommon, what monsters were best to score legendaries from, and more. Not only were the player replies interesting, they shed a lot of light on how items are generated in Diablo 3.One key factor is the item level and item type, and that’s almost entirely responsible for the extreme scarcity of some very desired items, including the Mempo of Twilight and the Witching Hour. It’s not that level 63 legendary quality helms drop that rarely, it’s that when one drops it’s got a 1/10 (at best) chance to be a Mempo.
A related issue is Set Item scarcity, and this one may surprise you. Set Items are *not* less likely to drop than legendary items. So why do you see so many more oranges than greens? Simple: 1) there are more total legendary items, and 2) so many of the set items are stacked up in the same slot and the same item level, essentially crowding each other out. For example, there are 4 set helms at lvl 63 and 5 set chest pieces at 63, and only one can drop each time. You’re actually 5x MORE likely to see a level 63 green chest piece than a lvl 63 orange chest piece, but that item drops so seldom that you’d have to play 5 years with 500% Magic Find to notice.
Click through for a synopsis of player replies to all of the survey questions, stats on which enemies (Elites vs. Trash Mobs vs. Gobbies vs. Chests) most often drop legendaries, plus interesting charts and graphs and data collections about this and more. Also see the Unofficial Official Diablo 3 Legendary Drop results thread, where we’ve got a community effort going to track and share drop rates of specific items, for the greater glory of statistical analysis.
Many different opinions on this question, though some general agreement. There were a good number of replies pointing to the 300-350% range as the start of good legendary drop rates, but it’s far from unanimous. Others said legendaries tip up once you’re past 250% MF, others not until 400% or more, and quite a few had no opinion. Solid minority of players say they do not have a high level paragon character, or play self found, or otherwise haven’t raised their MF high enough to notice any improvement yet.
BTW, the maximum possible Magic Find is easy to calculate since there’s a 300% hard cap on the MF from gear + paragon levels. Five Nephalem Valor stacks adds 75%, each level of Monster Power adds another 25%, and each additional player in a multiplayer game adds 10%. So a solo player at the MF cap gets 300% + 75% on MP0, 500% on MP5, and 625% on MP10. Adding three other players puts the maximium at 655% on MP10 with 4 players in the game.
The real issue of MF is shown in the graph to the right. It’s unscientific, but it does fairly represent the Legendary drops vs. MF experience of most players. No one outside of Blizzard has an exact plot of this curve, but most players feel that legendaries almost do not exist below 200% Magic Find, before skyrocketig up to “barely acceptable” drop rates past 300 or 400%. With gradual improvement above that.
Question #1 had kind of a long answer. Question 2 does not.
A couple of players said they thought they’d seen one at some point, but no one was positive. Everyone else was a solid and instant no. This issue was often mentioned in reply to later questions, and almost everyone thought that legendary drop rates needed to be greatly increased for new and low level characters.
A lot of agreement on this one. Orange Spirit Stones (monk helms), orange gloves (especially Frostburns), Ivory Tower shield, Andariel’s Helm, orange Barbarian belts, and orange Bracers. Orange wizard hats also drew considerable support, and by support I mean hate, especially for the Dark Mage’s Shade since it drops constantly in Inferno but is sub level 60 and thus doesn’t even salvage into a Fiery Brimstone.
Quite a few players also mentioned two-handed weapons, but no one was sure if they were really that common, or just seemed common since they’re all useless, save for the Skorn, which is only useless when it doesn’t spawn with Life Steal, huge DPS, and a great stat roll.
Data very useful to this and other questions came courtesy of qtk33 and Dana Scully two players who have been keeping detailed records of their legendary finds for months. These images and their data match very well with the anecdotal memories of players, regarding which are the most common and least common legendary items.
The odds of legendary items dropping are mainly determined by the item’s type, its ilvl, and if there are other legendary items of that type and the same level. (Really useful post by AurelianZ on this topic.) Players find a lot of unique shields and bracers and shoulders since there aren’t any set or class specific legendary items to split up the drops. That’s not the case with 3 other armor types, which all include some of the most valuable and rarest items in the game.
Helms: The Mempo of Twilight is widely reported to be amongst the rarest drops, and the fact that it is potentially the best helm in the game makes them seem all the scarcer. Why are they so uncommon, while level 62 Andariel’s Helms seem to drop every other game? Because lvl 63 legendaries are much less common that level 62 legendaries, and since when the game rolls a lvl 63 legendary quality helm it can choose from a Mempo or five different level 63 Set Helms, or three different legendary class helms: the Storm Crow, Visage of Giyua, and everyone’s least favorite orange hat, the Madstone. Thus Mempos have at best 1/10th the drop rate of other level 63 legendaries in item slots that don’t have competition from class legendaries and/or set items.Boots: The same pattern repeats, with ilvl 62 Fire Walkers dropping far more often than ilvl 63 Ice Climbers. Yes, lvl 63s are less often dropped, but the real problem is the four ilvl 63 set boots: Zuni’s, BK’s, IK’s, and Natalya’s. Those are all potentially very good boots, but if you’re wondering why you get 1 Ice Climber for every 15 Fire Walkers, there you go.
Belts: You know the story by now. Unlike with Helms and Boots there’s not some lvl 62 orange belt that everyone finds while desperately craving the ilvl 63 Witching Hour…. but at level 63 you’ve got four set belts, IK, BK, Inna’s, and Tal Rasha’s and even better, the Lamentation orange Mighty Belt… all of them cutting your odds of ever finding a Witching Hour.
All the above seems to be true, but it’s all based on observations and speculation since no one outside of Blizzard has access to the full game code. We did have that in Diablo 2 and clever coders in the community crawled through every bit of it. You can read about the whole topic in fantastic detail Diablo 2 Wiki Item Generation Tutorial, and that’s why there were all of those useful item drop calculators in Diablo 2 that could tell you the exact odds of any monster in the game dropping any particular item, item type, unique, etc.
Furthermore, fans discovered numerous bugs in the drop code in Diablo 2 — items that couldn’t appear at all or from some monsters, etc. And there were a ton of oddities and inbalances in drop rates due to how the Treasure Classes were set up in Diablo 2. Read the wiki articles for full details, but briefly each Treasure Class contained from 4 to 12 (or more) different item types, and when the game rolled an item drop it would first select the TC, and then an item from the TC (item quality was figures as well). Lower TCs were more likely than higher ones, and since items were weighted for drop odds within each TC (class-specific stuff was less likely to drop) this caused some specific item types to be almost impossible to find, while others were wildly common. The infamous green breast plate situation in Armor 18 is a good example.
Again, we don’t know exactly how item tables or treasure classes or the like are arranged in Diablo 3 since we don’t have access to the code. But from the extreme scarcity of some legendary items, all of which have multiple other legendaries of the same type and the same level, it’s obvious that there are some quirks and situations with poor design, such as only 2 legendary and no set helms at level 62, and 10 of them at level 63. The devs could go a long way towards improving the drop rates and equity of item distribution in Diablo 3 simply by fixing or adjusting some of these item generation oddities along with all the other Loot 2.0 changes.
A lot of different replies on this one and no clear consensus. Most legendaries are dropped by trash mobs or by Elites, but thanks to some players who have been keeping records for months and months, we have good data with big numbers. You can see the tables and charts (and contribute your own records to the group project) in our Legendary Drop Results thread, and here’s a quote from Dana Scully’s records:
As I keep track of every single legendary found since patch 1.05 I can supply somewhat accurate numbers. As of now I am at 871 legendaries found (all of them in HC mode).
423 (48.6%) were dropped by Elite Monsters 350 (40.2%) by white trash monsters 51 (5.9%) by Treasure Goblins 29 (3.3%) by Keywardens 14 (1.6%) out of regular chests/corpses etc. 4 (.5%) out of Resplendent Chests
These stats are from Hardcore, which matters since higher Monster Power increases the odds of trash mobs dropping bonus Rare+ quality items. Thus if you’re playing in Softcore on a very high MP level, you should find more legendaries from trash mobs than a typical Hardcore player would.
This data isn’t of huge use in targeting your play — it’s not like anyone was skipping Elites or golden chests or Gobbies since they thought their odds of dropping a legendary were too low — but it’s interesting to see the actual stats and compare them to your selective memory. For instance, I remember finding several legendaries from bodies and chests and even once from an armor rack… but they stick in my memory precisely since those clickables are so seldom the source of even a rare, much less a legendary.
Since the original Legendary Survey thread was posted I’ve been keeping records of every legendary I’ve found and it’s been eye-opening how different the actual finds are than what leaps to mind when I think about my legendary finds. I recommend keeping your own records; it’s fun to feel like you’ve gained some knowledge every time an orange or green drops (even when the item is worthless) and we’re compiling all the data via the Legendary Drop Rate thread, so you’ll be contributing to the effort.
This question makes no sense in light of what we’ve learned from the previous replies. Yes, lvl 63 items drop less often than level 62 items, which are less common than lvl 61 items, and so on. But what really matters is how many possible legendaries are in a given slot and item level, since they crowd each other out. Level 63 helms aren’t that much less common than level 62 helms, but there are 10 level 63 and only 2 level 62, so any particular level 63 helm is much less common than either lvl 62 helm.
I have to think that the developers didn’t really grasp the consequences of their itemization allocation during development. (Much as the D2 devs left their Treasure Classes very unbalanced until v1.10 smoothed them out quite a bit.) The D3 devs could and probably should have reduced the clumping, especially of item sets. Imagine if they’d spread out set helms and class-specific helms acros level 61-63; all of those hats would be considerably more common than they are now, and players would be happier with the item hunt.
This is about the only issue that didn’t see general agreement in the forum survey. Most (but not all) players feel that sets drop far too rarely, and players want to find set items more often. The irony is that set items actually drop at the same rate as legendary items… the reason you see so few of them is the way the item tables are set up, because most sets are stacked up with multiple pieces from different sets in the same slot at the same item level, which means they’re essentially crowding each other out.
I added the underline for emphasis, since that’s the key to understanding that sets aren’t actually less likely to drop, even though you see far fewer of them. The chest armor slot is an obvious example — if you roll a level 63 chest armor of legendary quality, there are 6 possible results: the legendary Tyrael’s Might, or one of 5 set pieces: BK, IK, Zuni’s, Inna’s, or Tal Rasha’s chest. Each of these items appears to have the same 1/6th chance to drop, but since they’re all in the same item level and slot, you’ll very seldom see any of them. And will probably see more Cindercoats (level 60 legendary) than all the level 63 set chest pieces combined.The one item type that breaks this pattern, and demonstrates that set items are actually no more uncommon that legendary items is jewelry. There are more total legendary rings and amulets that can drop in Inferno, but that’s because at level 61 there are 3 rings and 2 amulets, all legendary. Up at level 62 though there are actually more set items, with 3 set amulets vs 2 legendary amulets, and 5 set rings vs. 3 legendary rings. Thus players actually find more total green jewelry than orange jewelry at level 62, and the total number of set vs. legendary jewelry pieces is fairly close (though the level 61s drop more often than level 62s) so players find quite a bit of set jewelry.
The same thing would happen in other items slots if the set boots, belts, chest pieces, and helms were spread out over levels 61-63, instead of all stacked up at level 63, and if that simple change was made then players would instantly see far more set items dropping without the game changing anything else in the RNG that determines set item scarcity.
The last question received the most replies, with everyone offering an opinion.
Perhaps surprisingly, many players didn’t think that legendary drop rates needed to be improved… at the high end. Numerous players with a lot of Magic Find, 400% or more, said that legendaries dropped often enough now. That differs a great deal from players talking about new characters, or playing self found, or without high paragon levels who thought legendaries were far too scarce.
There was agreement on lower level legendaries though, with almost everyone of the opinion that legendary drop rates need to be *greatly* buffed for level 1-60. Especially since such items are generally of use for just 10 or 15 levels at most, by which time any random rare will outpoint them on stat quality.
I will echo the sentiments of most in this thread. I feel like I get plenty of legendaries with my plevel 68 on MP8 but when I was playing through the game with that character from 1-60, I only found 2 legs through 3 difficulty levels.
After lvl 60 I think the legendary drops are ok, dont need more, just more quality for most legendaries. But bellow 60 is just a joke, i lvled all the classes to 60, some after mp, so I was playing at least mp5, without any rush, and I just found 3 legendaries on average from 1 to 60. And I think that make boss running profitable somehow would make the game a bit more fun too.
There was general agreement on another issue, that legendaries were too poor in quality, if not quantity. The huge variation in their potential stats was often cited negatively, and a number of players actually want to see fewer legendaries drop, but with much better changes of being top quality. The devs haven’t spoken about frequency, but they have confirmed that part of Loot 2.0 will be considerable lifts to the floor on the quality of specific legendaries. No more 800 DPS 2H weapons; that sort of thing.
I will echo what others have said, in that the quality of legs/sets needs to be tightened up significantly. Especially when you consider that pretty soon we will be able to find ANY legendary at level 63. This will greatly diminish your chances of finding any one specific legendary, and so to me they have to at least make sure that when you finally do find that EF or whatever it is your heart desires, that there is more than a 1% chance its total garbage. Hitting the RNG jackpot once should be enough. You shouldn’t be forced to hit it for the item, and then for the affixes on the item, and then for the range of those affixes.
This is not the real question, IMO. The real problem with D3 is: why do I need to loot 100 “legendary” items before I get one that’s good? and I don’t mean good as in 1bil good, I mean good as in I can actually use it for a character… How, by any logic, do you take the concept of trash loot, magic loot, rare loot and legendary loot and make it ALL trash loot? What the hell is the point of even having “Legendary” items when 99.9% is useless garbage no better than a blue, like monk gear with STR, barb gear with INT, blah blah blah…
Another area of agreement was many players wanting runs back, or at least some specific monsters with better chances of dropping a legendary item. Random purples and Act Bosses were cited several times, though players specified these needed to be harder to kill or harder to reach to keep it from being totally unbalanced.
Runs in D2 were grossly exploited, but players enjoyed runs. The D2 ones were just (arguably) too short and easy, and certain bosses too good on their drop odds. Runs need to be better and targeted, but not so rewarding that they are the only sensible way to play.
The Diablo III devs had a good idea with changes to make more of each act profitably playable, but they went too far with flattening the drop rates. It’s fun to play when you’ve got something to target, and some monsters can drop different things than others. Players enjoy the KeyWardens even if they’re not really trying to make more Infernal Machines, just because the KWs are different and can drop something other enemies can’t.
I regret to report the idea I’ve been flogging for months, that the devs should add more/new/different dungeons, much deeper ones, with special +legendary drop purple bosses in them, was not mentioned by anyone. That seems the best and most obvious fix to me, as it would provide new content, varied content, and give players specific goals to seek for.
Thanks to everyone who posted a reply in the thread. They were informative in specific and on aggregate, and the data provided made clear some interesting details about how item drops work in Diablo 3 that I’d never seen posted before. It’s also interesting to see what players agree on and what they differ on, and how much that corresponds to individual play styles and time spent in-game. The thread also led to the ongoing Legendary Drop Rate results, which should add more useful info to the data we saw in this survey.
We don’t know how many of the main complaints will be addressed by blizzard in D3X and/or Loot 2.0. They have commented on some of the issues, such as tightening up the quality (raising the floor) to make total junk legendaries less common, but it remains to be seen how well they balance that, and how the odds of finding any particular legendary change when/if they implement scaling levels on legendary items, so you’ll be able to find scaled up versions of low level legs even in Inferno. (And vice versa, though since no one thinks legendaries drop often enough to matter for characters level 1-60, that’s almost irrelevant.)
The issue of re-introducing something like runs, or giving players some reason to want to kill Act Bosses, or to hunt down random purples is an open question.
And we really don’t know what’s going to happen to paragon levels and Magic Find levels or MF on gear in the future. Bliz has suggested we’ll see more customization from Paragon Levels, and that they’ll probably add more Paragon levels in the expansion… but would that mean higher Magic Find at the top end, or perhaps just a lift to the floor to boost MF early on, or even some kind of baked in Magic Find for all characters? And how might Paragon levels to an account, rather than just to individual characters affect that issue?
Clearly there’s a lot we don’t know yet and the months to come should reveal many interesting changes to Diablo III in the immediate and long term future.