Files
2nd/10_Wiki/Topics/AI_and_ML/상성 및 데미지 유형(Unit Counters & Damage Profiles).md
T
2026-05-10 22:08:15 +09:00

6.5 KiB
Raw Blame History

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack
id title category status canonical_id aliases duplicate_of source_trust_level confidence_score verification_status tags raw_sources last_reinforced github_commit tech_stack
wiki-2026-0508-상성-및-데미지-유형-unit-counters-damage 상성 및 데미지 유형(Unit Counters & Damage Profiles) 10_Wiki/Topics verified self
Unit Counters
Damage Profiles
RTS 상성
Rock-Paper-Scissors Combat
none A 0.9 applied
game-design
rts
combat-system
balance
2026-05-10 pending
language framework
csharp Unity-DOTS

상성 및 데미지 유형 (Unit Counters & Damage Profiles)

매 한 줄

"매 unit 의 매 다른 unit 에 대한 effective vs ineffective relationship". RTS 의 매 핵심 strategic depth — Dune II (1992) 의 light/heavy/infantry triad 에서 시작, StarCraft II 의 armor type × damage type matrix 로 evolution, 매 modern title (Stormgate 2026, AoE IV) 에서 매 attack-bonus 시스템으로 standardize.

매 핵심

매 두 축 — Damage Type × Armor Type

  • Damage Type: Normal, Concussive (anti-light), Explosive (anti-heavy), Piercing, Magic
  • Armor Type: Light, Medium, Heavy, Biological, Mechanical, Massive, Structure
  • Multiplier matrix: 0.5x (resist) ~ 1.5x (effective). 매 2.0x 의 X — 매 too swingy.
  • Bonus damage vs base damage: SC2 는 base + bonus_vs_ 분리. Balance lever.

매 Counter Triangle 패턴

  • Light counters Heavy counters Air counters Light — cyclical, no dominant.
  • 매 transitive counter (A>B>C>A) 는 매 strategic decision 강제. Asymmetric counter 는 매 rush meta 유발.
  • Soft counter (1.3x): unit 합성 권장. Hard counter (2.0x+): pure composition 강요 — 매 일반적으로 avoid.

매 응용

  1. Composition decision — scout 으로 enemy comp 보고 counter unit 생산.
  2. Map control — terrain × range 의 unit type 배치 (high ground siege, choke point splash).
  3. Tech tree pacing — counter unit 의 tier gate (T2 anti-air, T3 anti-massive).

💻 패턴

Damage type enum + armor matrix (Unity DOTS)

public enum DamageType : byte { Normal, Light, Heavy, Pierce, Magic, Siege }
public enum ArmorType : byte { Light, Medium, Heavy, Bio, Mech, Massive, Structure }

public struct DamageProfile : IComponentData {
    public DamageType type;
    public float baseDamage;
    public FixedList64Bytes<DamageBonus> bonuses;
}

public struct DamageBonus { public ArmorType vs; public float bonus; }

public struct ArmorProfile : IComponentData {
    public ArmorType type;
    public float armorValue;
}

public static float ResolveDamage(in DamageProfile dp, in ArmorProfile ap) {
    float bonus = 0f;
    for (int i = 0; i < dp.bonuses.Length; i++)
        if (dp.bonuses[i].vs == ap.type) { bonus = dp.bonuses[i].bonus; break; }
    return math.max(0.5f, (dp.baseDamage + bonus) - ap.armorValue);
}

Counter chart (data-driven JSON)

{
  "marauder": { "damage": 10, "type": "Heavy", "vs_armored": 10, "armor": "Bio" },
  "marine":   { "damage": 6,  "type": "Normal", "armor": "Bio" },
  "stalker":  { "damage": 13, "type": "Pierce", "vs_armored": 5, "armor": "Mech" },
  "zealot":   { "damage": 16, "type": "Normal", "armor": "Light" }
}

ECS damage system (burst-compiled)

[BurstCompile]
public partial struct DamageResolveSystem : ISystem {
    public void OnUpdate(ref SystemState state) {
        var armorLookup = SystemAPI.GetComponentLookup<ArmorProfile>(true);
        new DamageJob { armorLookup = armorLookup }
            .ScheduleParallel();
    }
}

[BurstCompile]
partial struct DamageJob : IJobEntity {
    [ReadOnly] public ComponentLookup<ArmorProfile> armorLookup;
    void Execute(in DamageProfile dp, in TargetEntity target, ref Health hp) {
        if (!armorLookup.TryGetComponent(target.value, out var ap)) return;
        hp.current -= ResolveDamage(dp, ap);
    }
}

Splash 응용 — counter via AoE shape

public struct SplashShape {
    public float innerRadius;   // 100% damage
    public float outerRadius;   // 25% damage
    public bool linear;          // siege tank line vs colossus disc
}

Balance simulation harness

import numpy as np

# armies: list of (count, dps_vs_type_dict, hp, armor_type)
def simulate(army_a, army_b, dt=0.1):
    while alive(army_a) and alive(army_b):
        damage_a_to_b = sum(c * dps[b_armor] for c, dps, _, _ in army_a for b_armor in types_in(army_b))
        # ... apply to softest target first (focus fire heuristic)
        tick(army_b, damage_a_to_b * dt)
        tick(army_a, damage_b_to_a * dt)
    return winner_margin(army_a, army_b)

매 결정 기준

상황 Approach
Symmetric RTS (SC2-like) Soft counter matrix, 1.3-1.5x multiplier
Asymmetric (faction-unique units) Per-faction counter chart, identity > balance
MOBA-style hero Stat-based counters (mobility, range, burst) — no explicit type
Auto-battler (TFT) Trait synergy + damage type secondary
Tower defense Armor-color hard counter (red = magic only) acceptable

기본값: SC2-style — Light/Armored/Massive armor tag + soft 1.3-1.5x bonus damage. 매 readable + balanceable.

🔗 Graph

🤖 LLM 활용

언제: damage matrix 의 first draft, balance hypothesis ("if zergling +1 damage, what changes"), counter chart 의 markdown 화. 언제 X: 매 actual playtested numbers — LLM 의 pure speculation. Sim harness + telemetry 가 ground truth.

안티패턴

  • Hard counter only (2x+): composition 강요 → boring "scout-and-react" gameplay.
  • Hidden multipliers: damage tooltip 에 안 보이면 매 player 의 frustration.
  • Symmetric mirror (모든 unit 의 동일 multiplier): counter 의 의미 X.
  • Too many damage types (8+): cognitive overload. 4-5 가 sweet spot.

🧪 검증 / 중복

  • Verified (Blizzard SC2 design archives, Stormgate 2026 design blog, Sirlin "Playing to Win").
  • 중복 candidate: 유닛 상성(Unit Counters) — 매 같은 folder 에 duplicate. 매 future merge 의 candidate.
  • 신뢰도 A.

🕓 Changelog

날짜 변경
2026-05-08 Phase 1
2026-05-10 Manual cleanup — RTS counter system + Unity DOTS damage resolution patterns