A damage-rate cap protects encounter structure from extreme burst while ordinary attacks still move the health bar. It estimates pressure with a recent-damage window or weapon-DPS model, then reduces later damage above a threshold. Define the window, curve, floor, decay, exception order, and feedback so stronger builds still feel effective.
Tavi lands an isolated 18-point sword hit at 0.92 s. With no recent damage in the window, all 18 points apply and Kern falls from 100 to 82 health.
Autoplay is disabled by reduced motion. Scrub across 0.34, 0.92, 1.72, 2.08, 2.43, 2.78, 3.13, 3.48, 4.48, 4.94, and 5.42 seconds to inspect the curve and reset.
Define the measured damage sample, window duration, threshold, attenuation curve, minimum multiplier, per-hit and damage-type exceptions, first-hit behavior, decay, phase resets, multiplayer aggregation, health application order, stable hit ids, rollback, and telemetry for raw, applied, and prevented damage.
Choose a rolling sum, weighted history, weapon burst-DPS estimate, or another explicit sample. Record which hits enter it, when they expire, whether players share a window, and whether the current hit is evaluated before or after it is added.
Map measured pressure to a multiplier with a documented threshold, curve, and nonzero floor. Keep identical input deterministic, clamp the result once, and decide the order of critical hits, armor, vulnerabilities, shields, and attenuation.
Make applied damage, health loss, and recovery from attenuation distinguishable. Test isolated hits, dense bursts, damage-over-time, multihit attacks, multiplayer focus fire, phase transitions, latency, duplicate events, and builds on both sides of the threshold.
If numbers, reactions, and health loss do not reveal attenuation, players read the system as a bug or dishonest scaling. Preserve meaningful gains below and above the threshold and expose the changed multiplier through feedback or encounter UI.
A near-zero floor, long memory, or reset that never arrives can erase valid play. Give the window a bounded lifetime, maintain a useful minimum contribution, and audit interactions that continuously refresh pressure.
For each hit, store hitId, sourceId, rawDamage, recentSample, threshold, multiplier, appliedDamage, preventedDamage, healthBefore, healthAfter, and resolution time. Replayed hit ids must not apply twice.
A rolling window can reward cadence and intentional pauses; a weapon-DPS model can equalize loadouts but may feel detached from moment-to-moment play. Pick the model that protects the encounter without invalidating the combat verbs you want players to express.
Track raw-to-applied ratio, time above threshold, damage prevented per source, kill time by build, phase skips, first-hit spikes, multiplayer concentration, floor uptime, and whether players recognize why their numbers changed.
One measured cut crosses the elemental’s shell cleanly. When Tavi crowds several blows together, the runes flare and spread the force across the stone; after a quiet beat their glow fades and the blade bites deeply again.
Combine health-bar motion, hit sound, impact weight, color, controller feedback, a protection aura, and an optional recent-damage meter. Offer a gentler curve, shorter window, higher floor, or explicit threshold indicator without changing which events qualify.
Armor reduces damage by a mostly stable rule, a per-hit cap clips one large hit, and a health gate stops progress at a fixed boundary. A damage-rate cap changes later hits according to recent or estimated pressure and relaxes when that pressure decays.
From 2:00, Hush’s second form demonstrates a long encounter protected from extreme burst. Its documented Damage Scaling tracks unmodified damage over the previous four seconds and lowers the multiplier toward a floor, making window length and minimum contribution central tuning values.
At 7:42, an extreme first-hit setup defeats Archon Boreal despite the encounter’s Damage Attenuation. The exception is useful design evidence: initialization and operation order matter as much as the curve, because a spike evaluated before enough pressure exists can bypass the intended pacing.
The boss fight begins at 11:00. The Fragmented One is documented with Damage Attenuation and a maximum DPS amount; the sustained firefight shows why designers must validate the cap across continuous weapons, high fire rates, status effects, and a long health-bar read rather than only single attacks.
Found an inaccuracy or have a suggestion for this page? Propose an edit on GitHub