Mechanics
Crates and loot
Last researched November 3, 20266 min read
Sources for this page
Applies to: ASA as of 2026-11-03 (DevKit UE 5.8.1); ASE where noted
ark.wiki.gg: Supply Crate, Item Quality (read 2026-11-03); my own reading of the 5.8.1 game's compiled code, checked against real saves and real crates (not published); my own modding guides (not published)
Confidence:AOfficial: Studio Wildcard, patch notes or in-game textBRead from the game or DevKit filesCInferred by comparisonDCommunity-measuredEUnverified
1. Crate tiers
Supply drops (beacons) come in colors with a minimum player level. The levels are the ones the DevKit 5.8.1 crates
need to open (SupplyCrate_Level03 to SupplyCrate_Level60, their _ScorchedEarth copies and Aberration's
SupplyCrate_Cave_Aberration_ and _Aberrant_Surface crates; conf. B), and every cell matches the wiki's
Supply Crate table (read 2026-11-03):
| Color | The Island / The Center / Ragnarok | Scorched Earth | Aberration |
|---|---|---|---|
| White | 3 | 3 | 10 (cave crate only) |
| Green | 15 | 15 | 25 (cave crate only) |
| Blue | 25 | 30 | 35 |
| Purple | 35 | 45 | 50 |
| Yellow | 45 | 55 | 65 |
| Red | 60 | 70 | 80 |
Most ringed (double) crates roll better quality than plain ones, and most roll more sets (DevKit 5.8.1, conf. B). On The Island a ringed crate rolls 4 sets at 150% to 200% quality; a plain one rolls 2 to 3 sets (white) or 2 to 4 sets at the default 100%. Aberration's ringed cave crates roll 2 sets (3 at level 80) at 200% to 250%, against 1 to 3 sets for plain cave crates. Aberration's ringed surface crates roll the same quality as the plain ones (150% to 310% at level 50, 165% to 370% at level 80), and only the level 80 one rolls more sets (3 to 4 against 2 to 4). Scorched Earth's ringed crates roll up to 300%; from level 3 to 30 they roll as many sets as the plain crates, and from level 45 they roll 5 or 6.
Deep Sea crates need level 80 (SupplyCrate_OceanInstant, conf. B) and usually hold Ramshackle or better.
2. How a crate rolls
Decoded from UPrimalInventoryComponent::GenerateCrateItems (conf. B). r is a uniform random number from 0 to 1, and a weighted pick takes the first index whose
running sum reaches r × the total.
- Number of sets:
MinItemSets + round((Max − Min) × r^NumItemSetsPower), unless quantity weights override it. - Each set: picked by the crate's own inline
SetWeight. If the set names anItemSetOverride, the override's contents (entries, item counts, power, without-replacement flag) replace the set's, but the crate's weight still decides how often it is chosen. - Items per set:
MinNumItems + round(span × r^NumItemsPower). - Each item: pick an entry by
EntryWeight; give nothing unless r ≤ChanceToActuallyGiveItem; then pick the item class byItemsWeights. It is a blueprint when r <ChanceToBeBlueprintOverride. - Quantity is a number of picks.
qty = Min + round((Max − Min) × r^QuantityPower)(times a multiplier the simulator does not model), and the item is re-picked for each unit, so a quantity of 5 can give five different items. OnlybApplyQuantityToSingleItemgives one item class as a stack. - Without replacement: an entry is removed after use when the set says so; a set is removed when the crate says so.
Inheritance: an override set that leaves a field unset takes its parent class's value, not the struct
default. The main ASA loot sets inherit 1 to 2 items at power 0.4 from LootItemSet_SupplyDrop_ASA_Base; structure
sets inherit 3 to 5 without replacement from their base while the child stores 1 to 3 (conf. B).
Checked against real crates (conf. B)
On 2026-10-06 I opened 25 summoned Aberration cave crates (level 35) and 61 surface crates and counted every slot:
| Crate | Real mean rolls | Model with parent inheritance | Plus quantity as picks |
|---|---|---|---|
| Cave, level 35 (25 crates) | 3.04 (3.12 slots) | 2.66 | 2.77 |
| Surface, level 35 (61 crates) | 5.67 | 5.33 | 5.43 |
The fit's log likelihood went from −294 to −163 once parent inheritance and quantity-as-picks were modeled (z 1.1). A residual remains: 6 of 25 cave crates had 5 or more items where the model expects about 1.
Facts the test settled:
- Beacon contents are never saved. A save made after summoning held no crates; an official Aberration save had only the 6 level-placed artifact crates. Rolls exist only in memory, so inspect them in the inventory UI.
- Stacks over the maximum split into slots (139 metal shows as 100 + 39).
- The Tek Replicator, Cryofridge and empty Cryopod that sit in every crate's folders are craftable engrams, the inventory's default items, not loot.
3. Quality: from crate to item
The wiki (conf. A)
(Item Quality, read 2026-11-03; the page is not split by game.)
- The loot system gives an arbitrary quality AQ.
- Each stat rolls RQ uniformly from 0 to AQ × RO (the randomizer range override).
- Per-stat quality is (RQ² / (AQ × RO)) × RV; their mean is the item rating, compared with the tier thresholds.
- Stat value = IV + IV × MS × DM + IV × MS × RM × RQ.
- Crafting cost = BC + BC × IR × RS × CM × 0.3.
| Tier | Threshold | Crafting XP × | Resources × |
|---|---|---|---|
| Primitive | 1 | 1 | 1 |
| Ramshackle | 1.25 | 2 | 1.33 |
| Apprentice | 2.5 | 3 | 1.67 |
| Journeyman | 4.5 | 4 | 2 |
| Mastercraft | 7 | 5 | 2.5 |
| Ascendant | 10 | 6 | 3.5 |
Caps it lists: most weapons 755.3% damage, Tek weapons 296.6%; saddles 325% armor (290% on tanky creatures), the saddle figures marked as needing testing (so conf. E).
Decoded (conf. B)
The full chain (fit to 104 tiered items from 86 real crates):
- Crate multiplier. A crate multiplier M above 1 becomes (M − 1) × 1.2 + 1 (
AboveOneExtraQualityMultiplier, native 1.2), then × the server'sSupplyCrateLootQualityMultiplier(and any event crate modifiers). - Entry roll. q = lerp(Min × entry
MinQuality, Max × entryMaxQuality, r^QualityPower). - Item.
InitNewItem, only for items that use stats: q × ((DifficultyValue − 1) × 0.5 + 1). For each used stat, f = r, floored at min(MinRandomQuality, q) / q, and the rating gains f × q ×RatingValueMultiplier. Weight adds −0.3 × f × q and is not counted. The item rating is the mean. - Tier. The rating must be strictly above a threshold: 0, 1.25, 2.5, 4.5, 7, 10 (read from the game data asset, since the native table is empty).
DifficultyValue is lerp(0.5, 4.0, DifficultyOffset); OverrideOfficialDifficulty wins, and MaxDifficulty forces 5.
The server multiplier shows clearly in real rolls: my single-player Game.ini had
SupplyCrateLootQualityMultiplier=2, and the fit preferred it decisively (log likelihood −23 against −167 at 1).
Where they disagree
| Point | Wiki | Decoded |
|---|---|---|
| Primitive threshold | 1 | 0; every tier needs a rating strictly above its threshold |
| Per-stat roll | RQ² / (AQ × RO): a squared uniform roll, which pulls ratings low | f × q with f uniform, floored by the entry's MinRandomQuality |
| Difficulty | not mentioned | quality × ((DV − 1) × 0.5 + 1) |
| Stat value | has a default-modifier term (DM) | no DM term in the native getter (below) |
The decoded version predicts real crates well but over-predicts stat items on surface crates by about 18% (102 against 84), an open draw-count question.
4. The item stat formula
Measured by calling the native UPrimalItem::GetItemStatModifier on items with set rolls (conf. B):
x = v × StateModifierScale × RandomizerRangeMultiplier (v = the raw roll, 0..65535)
percent stat: 1 + x
flat stat: InitialValueConstant × (1 + x)
709 of 714 values were exact across 289 classes. The five misses override the stat in Blueprint (Mek backpacks, a saddle, the Scout remote), so the game's own getter is the one to trust.
| # | Item stat | # | Item stat |
|---|---|---|---|
| 0 | generic quality | 4 | clip ammo |
| 1 | armor | 5 | hypothermal insulation |
| 2 | max durability | 6 | weight |
| 3 | weapon damage % | 7 | hyperthermal insulation |
- Durability is saved as a fraction of the maximum; above 1.0 is a real over-max state.
- Weapon damage does not live on the item: instant weapons use the template's hit damage × traces per shot,
projectile weapons the projectile's direct plus explosion damage, melee
MeleeDamageAmount(Damage and armor).
5. Spoilage
Spoil time = the item's SpoilingTime × the container's multiplier (conf. B):
| Container | Multiplier |
|---|---|
| Preserving Bin | ×10 |
| Industrial Preserving Bin (Steampunk Ascent) | ×10 for food, ×15 for raw and cooked meat and prime meat |
| Feeding Trough | ×4 |
| Tek Trough | ×100 for food, ×10,000 for iced water containers |
| Refrigerator | ×100 for food, ×10,000 for iced water containers, ×18 for Shocking Tranquilizer Darts |
Powered and switched containers apply nothing while off or unpowered. Preserving Salt, saddle containers and a player skill add their own multipliers.
6. Other loot
Treasure maps, buried chests and Tides of Fortune bottles share one system. My own guide to it, which I keep private, has the full flow (how the spot is chosen, why copies cannot dig, the bottle chain's odds sitting on the shovel).