Guide:Sparkly Chests v1.10, by Quickdeath
Sparkly chests are special glowing chests that may drop multiple items, including multiple set, rare, or unique items. Not all chests that drop like a "sparkly chest" are actually sparkly. Their drop odds are calculated in unusual fashion, and magic find does not affect their gifts.
Strategy guide info:
- Version: v1.10+
- Game type: PvM
- Character: All
- Author: Quickdeath
A Guide to Sparkly Chests – Unique and Set Items
Sparkly Chests are among the most complicated and wonderful items in Diablo 2 V.10/V1.11. They are fixed objects in the game –they do not randomly appear. They have very complicated algorithms for dropping –so complicated that they are not in the ATMA calculator. They usually drop a large number of items. They are unaffected by your character’s MF rating but are capable of dropping unique and set items – and are quite capable of multiple unique or multiple set drops
As I have dug into the implications of the algorithms for Sparkly Chests used to determine the probabilities of which objects drop, I have discovered a number of surprising things. But before I mention those, let me acknowledge the work of others. Thanks to TheJarulf, who made some of the major discoveries about Sparkly Chest drops and who graciously answered my questions in the Statistics Forum. Also, special recognition to Ravanal for his early work, as well as Urlik, Hrus, and Warrior of Light for their very helpful guides.
Okay, now, what have I learned about Sparkly Chests?
1. First, there is one particular Sparkly Chest(SC) in Hell that is a statistically rare source of TC87 uniques and set items. 2. SCs are far more sensitive to Player’s Setting that any monster. 3. The Sparkly Chest drop rate of uniques is relatively poor in the early acts of Normal difficulty, is much higher in Hell and highest in Nightmare. 4. SCs are the only objects in the game that have a chance of dropping 7 uniques or 7 set items – and the four with the best chance to drop 7 uniques are in Act 5 Nightmare: Abaddon, Glacial Trail, the Pit of Acheron and Drifter Cavern.
Basic Backgound Sparkly Chests have six basic cases for how they drop items (this was discovered by TheJarulf). Only the first two are of interest for Set and Unique drop odds:
Case A, where the intended quality of item is UNIQUE, happens 2 times out of 100, when you open an SC. When this case occurs, the sparkly chest will drop anywhere from zero to 7 uniques. In a Case A drop, all items will be either unique or magical or less; no Set items will drop.
Case B, where the intended quality of item is “Set”, happens 5 times out of 100, when you open an SC. When this case occurs, the sparkly chest will drop anywhere from zero to 7 Set Items. In a Case B drop, all items will be either set or magical or less; no Unique items will drop.
There are four other kinds of drops, where the intended quality of items dropped are rare or magical, or where there is no intended quality. These comprise the other 93 out of 100 cases when you open an SC, and except for a very rare circumstance, these drop cases will not generate unique or set items. This guide will ignore these other cases, and concentrate on Case A and Case B drops, which occur 7% of the time.
What happens under Case A or Case B? Whenever Case A (intended quality = Uniques) or Case B (intended quality = Sets) occurs, the Chest will pick four times using a set of tables. It will attempt to upgrade every item it picks to either unique (Case A) or SET (Case B.) Often the item picked cannot be upgraded because - No item got dropped (the code randomly picked “No Drop”) - The item dropped has no unique or set version (it might be gold, a health potion, a scroll or even a dimensional blade!) - The item dropped does have a Unique or Set version, but the Unique or Set version cannot be generated because either the area level or player’s character level is lower then the Unique/Set item’s qlvl. If the first item dropped (not a no-drop, but an actual item) is the intended quality (Unique or Set) then the code finishes picking four items to drop, upgrades them all to Unique/Set as possible, and stops. In this instance, the yield is 1 to 4 Unique or Set items.
However, if the first item dropped is not the intended quality (which happens about 85% of the time) then the code finishes the sequence of four picks, upgrading them as possible, and picks a second batch of another four items to drop. The same rules apply to this second batch of picks –they will be upgraded to Uniques or Sets, as possible. If the first item dropped from the second batch is of the intended quality then the second batch terminates the sequence. If the first item of the 2nd batch is not the intended quality, the chest then goes into a consolation prize mode and starts dropping items 4 at a time with quality magic or lower. Anywhere from 4 to 44 more items can then drop. Except for one very rare case, no item after the first 8 will be upgraded to the intended quality, and MF does not affect the subsequent drops.
First stage of picking items. At the top level of logic in Cases A and B, are these very important pick odds:
Equipment (armor,weapons) Frequency = 10 Good (Gems, rings, amulets, jewels, charms) Frequency = 2 Gold Frequency = 15 Junk Frequency = 15 “No Drop” Frequency = 2- 100 (depends on number of players).
The no drop frequency Table is seen to the right.
“Equipment” and “Good” are categories that can drop the intended qualities of Unique or Set. Without the “No Drop” factor, Equipment and Good will be picked 12 out of 42 times; but including the “No Drop” factor means that “Equipment” and “Good” will only be picked 12 out of 44-142 times.
This is an enormous difference. At 4 players, which is the highest player mode achievable in Single Player (that is, in players7 or players8 mode) the Equipment and Good categories will be picked 12 out of 55 times, 21.8% of the time. With 1 player, the desired Equipment and Good categories will be picked 12 out of 142 times, or 8.5%. Thus, in a single player mode, shifting your players setting from players1 to players7 will make a difference of 2.5 X in the probability of Unique and Set items dropping. This is far greater than the analogous effect on Act Bosses, where the No Drop frequency for Players1 is much lower (for Mephisto it is 15).
Whether or not you choose to change from Player1 to Player7 before opening a Sparkly Chest is a matter of personal opinion about whether this is cheesy or not. I am just informing you that your player setting makes a huge difference in the odds of getting items to drop from a chest.
For most of the rest of the guide, I will assume that the Sparkly Chests are opened at players7 or players8 (i.e., simulating four players), either because you’re playing at that level, or because you’ve shifted up to that level before opening the chest.
Second Stage of Picking Items. Given Case A or B (at 2% and 5% probability, respectively), the code picks one of the above categories. In 10 of 55 it picks equipment, in 2 of 55 it picks “Good,” in the other 43 of 55 cases it picks ‘Gold,” “No Drop” or Junk and will not drop a unique/set.
If the random number generator picks “Good”, the routine then picks further:
Gems, with frequency that is usually 46/130, Runes, usually at 14/130 or Jewelry, at 60/130
Gems (e.g., topazes, diamonds, skulls, etc) and Runes have no Unique or Set versions so the selection of Gems or Runes is a dead end. Jewelry, however, is subdivided as follows:
Rings, 8 of 20. Amulets, 4 of 20 Jewels, 2 of 20 Grand Charms, 2 of 20 Large Charms, 2 of 20 Small Charms, 2 of 20
Note that Rings and Amulets can be upgraded to Sets and Uniques, and that Jewels can be upgraded to unique Rainbow Facets (for Sparkly Chest in areas where alvl =85.) Grand Charms can be upgraded to the unique Gheed’s Fortune, wherever the Chest is in an area with alvl =70 or higher – basically anywhere in Hell difficulty.
If, in the first step, the code instead picks Equipment (rather than Good), the code then picks an armor or weapons Treasure Chest (TC) category through a bewildering set of table calls. In reality, these table calls were the programmers’ way of writing software that can easily be updated to include new items or new, higher-level TCs. Upon closer inspection, each chest in the game is assigned a set of deterministic probabilities as to which equipment item is called.
As an example, whenever the Equipment category is picked, the TC Weap60 will be chosen by the Hell Cave L2 Sparkly Chest 2.44% of the time while the Ancient Tunnels Sparkly chest will pick it 2.57% of the time. Not much difference in those numbers? Let’s look closer.
TC Weap60 has 9 base weapons in it, only 3 of them are upgradeable to unique and none are upgradeable to Set. The base weapons and their unique version are:
At frequency level of 3/19: - Blade Bow (qlvl 60) - Bone Knife (qlvl 58); Unique: Wizardspike (qlvl 69) - Hyperion Spear (qlvl 58); Unique: Arioc’s Needle (qlvl 85) - Legend Sword - Ogre Axe (qlvl 60); Unique: Bonehew (qlvl)72
At frequency level of 2/19: - Suwayyah (qlvl 59)
At frequency level of 1/19: - Heavenly Stone (qlvl 59) - Walking Stick (qlvl 58).
If the intended quality is Set(Case B) then we are out of luck because none of the items in weap60 can be upgraded to Set. If the intended quality is Unique (Case A) then 9 times out of 19 a Bone Knife, Hyperion Spear or Ogre Axe will be picked and they are all upgradeable to Unique. Assuming the virtuous player has opened the Sparkly Chest using a character with a clvl=85 or higher, and the chest is, say, the Ancient Tunnels Sparkly Chest (alvl=85), then those three base items are eligible to be upgraded to Wizardspike, Arioc’s Needle or Bonehew. But if the player has opened the Cave L2 (H) chest, where the alvl is only 79, then the Hyperion Spear cannot be upgraded to the qlvl 85 Arioc’s Needle (it will comvert instead to a rare hyperion spear with 3x durability). The Cave L2 (H) SC can only yield a Wizardspike or Bonehew (at a frequency of 6/19) when weap60 is picked. Thus, the TC weap60 category can have different outcomes, depending upon the alvl of the chest from which it was summoned.
So that’s the methodology, as simply as I can explain it. I’ve put together a large spreadsheet and have calculated the probability of every Sparkly Chest in the game calling every TC, and then I’ve figured out which Uniques and Set items are upgradeable from base items by which Chests. So, lets get to some results.
Super Chests and their yield of Uniques/Sets.
First, let me remind everyone that not all chests that sparkle are “Sparkly Chests.” Some sparkling chests are “Quest Chests.” These are chests that yield Quest items and they have different rules for dropping items. Quest Chests include, among others, the Maggot Lair chest, The Spider Cavern Chest, the Kurast Lvl 2 chest, the Ruined Temple chest (Lam Esen’s Tome), etc. True “Sparkly Chests” do not yield a quest item.
There are 24 Sparkly Chests within each difficulty level –or a total of 72 on the path to Hell Baal. There are five chests with identical characteristics in the False Tal Rasha’s Tombs. The Great Marsh has two Sparkly Chests with identical characteristics. The other 17 chests are each individualistic in their drop odds as they are encountered on Normal, Nightmare (NM) and Hell (H) difficulty levels.
Below is a list of the Sparkly Chests in Normal difficulty and the average number of uniques/sets dropped each time you open the chest. “No-drop” odds used in these calculations are those for 4 players (players7 mode), it is also assumed that if the first four picks are all “No Drop” that the sequence diverts into another case sequence and no Unique items are dropped at all. It was further assumed that the character opening the chest has a clvl equal to or higher then the area level
Normal Sparkly Chests and their Unique/Set Drop Odds
The Normal Cave L2 Sparkly Chest cannot drop uniques or set items; its area level is too low. In fact, all the Act 1 Normal Chests are poor fishing grounds for Uniques and Set Items. This probably creates an impression with many players that Sparklies don’t yield much. Avoid getting confused on the Kurast Sewers Sparkly Chest. There is a Quest Chest on L2, but randomly located somewhere on Kurast sewers Level 1 is a true Sparkly Chest.
Act 3 and Act 5 Nightmare Sparkly Chests offer the best odds of getting an S/U to drop. Here are the Nightmare Sparkly Chests:
Nightmare Sparkly Chests and their Unique/Set Drop Odds
Sparkly Chests in Hell will draw from higher TCs that cause the drops odds to go up slightly. But the higher TCs and area levels mean that better objects can drop. Here are the Hell Sparkly chests.
Hell Sparkly Chests and their Unique/Set Drop Odds
Note that it is the relatively ignored Drifter Cavern that has a Sparkly Chest with a small chance of dropping from TC87. However, the Chest can only drop items that have a qlvl of 84 or less. Fortunately, this turns out to be most of the TC87, namely:
Deathcleaver (Qlvl=78) Darkforge Spawn (72) Astreon’s Ironward (68) Griswold’s Redemption (44) Griswold’s Valor (44) Griswold’s Honor (44) Bul Kathos’ Sacred Charge (50) Bul Kathos’ Tribal Guardian (50) Griffon’s Eye (84) M’Avina’s True Sight (21) Death’s Fathom (81) Stormspire (78) Executioner’s Justice (83) Windforce (80) Ghostflame (70) Shadowdancer (79) Templar’s Might (82) IK Soul Cage (37) Natlaya’s Mark (22) Earthshifter (77) Death’s Web (74) Steelpillar (77) Gargoyle’s Bite (78)
Why is the Drifter Cavern Chest different than all the other Sparkly Chests? Because it is the only Chest in Act 5 Hell with an area level of 84, thereby qualifying it to drop from the TC category called Act5 (H) ChestB. Act5 (H) ChestB has a small chance of choosing either WEAP87 or ARMO87 to drop from.
Is Drifter’s Cavern Sparkly Chest a good source of TC87s? No, not really. Each time you pop the DC Chest, the cumulative TC87 Set/Unique drop odds are only about 1:13,680.
Below is a list of the Sparkly Chests and the average number of uniques dropped each time a Case A drop is picked (odds, 1;50). “No-drop” odds used in these calculations are those for 4 players (players7 mode), it is also assumed that if the first four picks are all “No Drop” that the sequence diverts into another case sequence and no Unique items are dropped at all. It was further assumed that the character opening the chest has a clvl equal to or greater then the items that can drop.
Mean Number of Uniques per Case A drop.
Does this mean that once in 50 times you open a Sparkly Chest, that at least one unique will drop? Heck, no. Let’s check out the frequency of zero, 1 and multiple uniques dropping for the Mausoleum SC in Normal, NM and Hell difficulties.
Note that a drop of Seven Unique Items can drop only in this sequence: (non-U),U,U,U __ U,U,U,U That is, the first item dropped in the first batch of four picks must be a non-unique, thus triggering the 2nd batch of 4 picks. The remaining picks must all then be unique – a very unlikely scenario.
Frequency of Multiple Uniques dropping, Mausoleum SC, Case A Drop @ 4 players
Thus, the drop odds for a 7 Unique drop from a Hell Mausoleum Sparkly Chest on players7 (4 players) are 1:1,459.357,479 – more than one in a billion! However, drops of 2-5 Uniques are considerably more likely; I’ve personally had a drop of three uniques from an SC –something you would see from the Hell Mausoleum chest every 800 runs or so.
Just to show the leverage of the players setting, if a player was running Hell Mausoleum SC in a multiplayer game with 8 players located nearby, the drop odds of a 7-Unique would be about 1: 306,049,446 -or 5 times more often than on Single Player P7. In contrast, on Players1 setting, one would have odds for a 7-unique drop of more than 1: 1.1 Trillion –a factor of almost 700 above the 4-player case! The only silver lining is that, after 1.1 Trillion runs, you would probably drop a lot of Zod Runes as well!