Skip to main content
UMI gives you multiple knobs to control rarity:
  • Global rarity weights on the loot table
  • Per-item rarity overrides in FLootEntry
  • Level-based tuning via ItemLevel or actor properties

​
Global Rarity Weights

Configure in your LootTable: Example:
  • Item.Rarity.Common → 50.0
  • Item.Rarity.Uncommon → 30.0
  • Item.Rarity.Rare → 15.0
  • Item.Rarity.Epic → 4.0
  • Item.Rarity.Legendary → 1.0
Weights are relative; you don’t need them to add up to 100.

​
Item-Level Scaling (Example Pattern)

Higher-level loot sources can favor better rarities. Example (conceptually):
  • Level 1 Chest:
    • Common: 60%
    • Uncommon: 30%
    • Rare: 10%
  • Level 10 Chest:
    • Common: 30%
    • Uncommon: 35%
    • Rare: 25%
    • Epic: 9%
    • Legendary: 1%
Implementation is game-specific. One approach: Blueprint Example 
On Generate Loot (or Post-Process Result)
   ↓
Get ItemLevel from LootComponent
   ↓
Select rarity weight profile based on level
   ↓
Override Global Rarity Weights
   on the LootTable or via a custom rarity resolver
You can also implement a C++ helper that:
  • Takes ItemLevel
  • Returns weighted rarity
  • Applies that to each FLootedItem::RolledRarity

​
Per-Item Rarity Override

Force a specific rarity for an entry: Example entry:
  • Item Data: DA_LegendarySword
  • Drop Weight: 0.1
  • Quantity: 1
  • Rarity Override: Item.Rarity.Legendary
This sword will always roll as Legendary, regardless of global weights.

​
Stat Scaling

Rarity and level can drive stat multipliers.

​
Stat Scaling Rules (Conceptual)

In the LootTable, you can define “stat scaling rules” such as:
  • Stat Tag: Stat.Attack
  • Rarity Scalars:
    • Common: 1.0
    • Uncommon: 1.2
    • Rare: 1.5
    • Epic: 2.0
    • Legendary: 3.0
Example (for a 10 base Attack sword):
  • Common: 10 × 1.0 = 10
  • Uncommon: 10 × 1.2 = 12
  • Rare: 10 × 1.5 = 15
  • Epic: 10 × 2.0 = 20
  • Legendary: 10 × 3.0 = 30
Exactly how you apply those multipliers (GAS effects, item stat struct, etc.) is left flexible on purpose.