Guide

Game power crush armor movement systems explained

Harbor Brawl's midscreen meta was stuck: Sentinel zoners fired projectiles from max range, then clipped every drive rush approach with standing jab. Stationary super armor heavies lost to fireball plus check before they activated. Ranked data showed zoner win rate at 58% in neutral-heavy matchups and average round length stretching past 42 seconds. Players called it “walk backward simulator.”

The fix was a dedicated power crush layer: a metered, forward-moving attack with one-hit super armor during startup, an explicit crush-high hurtbox phase that ducks under standing normals and projectiles tagged as mid/high, and a long recovery window punishable by low-profile sweeps. After tuning travel distance, armor frame count, and a 1.5-bar meter cost, zoner neutral win rate fell from 58% to 42% and average approach success rose 31% — without increasing rushdown damage. This guide covers power crush taxonomy, detection pipelines, crush-high vs armor interactions, low punish and throw counterplay, the Harbor Brawl refactor, a technique decision table versus stationary armor and plain dashes, pitfalls, and a production checklist.

What power crush is (and what it is not)

A power crush is an advancing move that combines three properties in one package: forward displacement, limited super armor during a defined startup window, and a hurtbox phase that crushes highs — standing attacks and mid-height projectiles whiff or pass overhead while the move continues. It is the footsies tool that answers “I cannot reach them because jab checks stop every dash.”

Not the same as crush counter

Crush counters reward startup priority trades on slow normals — the defender's active frames lose to your startup intercept. Power crush does not require winning a trade; it forces the opponent's standing check to miss via hurtbox height and armor while you close distance. Document both systems separately in movelists.

Not the same as low crush

Low crush ducks under highs without necessarily granting armor. Power crush often includes a crush-high phase and armor — but the armor can break on lows while the crush-high only evades mids. Mixing the tags produces unreadable interactions.

Not the same as hyper armor command normals

Stationary armor heavies absorb hits in place. Power crush adds movement: the attacker closes gap while armored. That movement is what makes it a neutral system rather than a trade tool.

Power crush state machine

Production teams implement power crush as a finite-state machine with explicit frame budgets. A typical 24-frame power crush might look like:

Frames 1–3: Vulnerable startup — no armor, standing hurtbox. Opponents can interrupt with jab or low.
Frames 4–12: Power crush active — one-hit super armor, crush-high hurtbox, forward velocity 2.8 units/frame, projectile mid-tag immunity.
Frames 13–16: Active hitbox — armor ends; strike connects or whiffs.
Frames 17–24: Recovery — standing hurtbox, launchable, throwable, minus on block if blocked during active.

The vulnerable pre-armor frames are non-negotiable. Zero-frame armor on a gap closer deletes the defender's check option and feels oppressive online. Budget 2–4 frames where a standing jab still wins — that creates the mindgame: “Do I check, low, or backdash?”

Armor hit budget

Most power crushes absorb exactly one hit during the crush window. Multi-hit strings break armor on the second active frame unless you implement hyper-armor tiers (rare for power crush). Chip damage from blocked attacks during armor typically still applies — otherwise defenders hold block and ignore the approach entirely.

Travel cap

Cap total forward displacement (often 1.5–2.5 body widths). Uncapped slide power crushes corner opponents from midscreen in one input and collapse neutral geometry. Pair travel caps with pushback rules so corner loops do not stack.

Crush-high detection and projectile tagging

Crush-high is a hurtbox-height override, not a hitbox priority flag. During crush frames, the defender's standing jab active frames pass above your lowered hurtbox capsule. Implementation requires:

Per-frame hurtbox height band (e.g. crouch or low override on a forward-moving body).
Hitbox height tags on opponent attacks: mid, high, low, overhead.
Projectile collision uses the same height tags — mid fireballs whiff; low-skimming projectiles still connect.

Animation must match: if the sprite still stands tall while hurtbox crushes, players report bugs even when math is correct. Add a visible crouch or shoulder roll silhouette during crush frames.

What still beats crush-high

Low attacks — crouching kick, sweep, low-profile normal during vulnerable startup or recovery.
Throws — armor does not beat throws unless explicitly tagged throw-invulnerable (avoid on power crush).
True overheads if tagged overhead and aimed at standing guard — crush-high does not duck under jump-in overheads aimed at head height.
Launch punishes on recovery — the primary risk/reward lever after a blocked or whiffed power crush.

Meter economy and neutral role

Ungated power crush spam turns neutral into armor roulette. Common economy patterns:

Meter cost: 1–2 bars per attempt; refund partial meter on hit to reward successful approach.
Cooldown: once per neutral reset or per round — used in games that want power crush as a comeback tool only.
Character allocation: 1–2 power crush tools per rushdown archetype; zoners get zero or a defensive reverse power crush with shorter travel.

Power crush sits between plain dash (beats nothing, fastest) and invincible reversal (beats everything briefly, slowest). It is the “I read your check” option in the neutral triangle: dash loses to check, power crush beats check, low beats power crush startup.

Harbor Brawl refactor: before and after

Before the patch, rushdown characters had only drive rush and stationary armor heavy as gap closers. Telemetry showed:

Drive rush approach interrupted by jab: 67% of attempts within 3 body lengths.
Armor heavy activated after jab already landed: 41% of attempts.
Average neutral engagement distance: 4.2 body widths (max range).

The team added a universal 4HP+K power crush (22 frames, 1.5-bar cost) with 3-frame vulnerable startup, 9-frame crush-high armor window, and 8-frame launchable recovery. Post-patch:

Jab check success vs power crush route: 38% (down from 67% on drive rush alone).
Low punish rate on whiffed power crush: 44% (intentional risk lever).
Neutral engagement distance: 2.8 body widths.
Zoner neutral win rate: 58% → 42%.

Key tuning insight: lowering recovery by 2 frames made low punishes impossible at online delay; keeping 8-frame recovery preserved the defender's reward for reading armor.

Technique decision table: power crush vs alternatives

Approach tool	Beats standing check	Loses to	Best when
Plain dash / run	Nothing inherent	Jab, low, projectile	Already at close range; meter empty
Drive rush (cancel)	Nothing; fastest gap	Check, anti-drive tools	Frame trap follow-up; corner push
Stationary super armor	One hit in place	Low, throw, multi-hit	Absorb one poke at current spacing
Power crush	Check + mid projectile (crush-high + armor)	Low, throw, recovery launch	Midscreen vs check-happy defender
Invincible reversal	Active frames fully	Block, bait, whiff punish	Defensive reversal; high commitment
Empty jump / bait	Forces whiff	Anti-air, fast check	Conditioning before real approach

Common pitfalls

Zero vulnerable startup. Removes the low-check mindgame; defenders always crouch and neutral flattens the other direction.
Multi-hit armor on gap closer. Strings cannot answer; zoners cannot hold space.
Throw immunity during crush. Deletes shimmy and throw counterplay unless intentional for a boss move.
Uncapped travel. One input crosses screen into infinite corner pressure.
Confusing power crush with crush counter in UI. Different systems; use distinct icons and tutorial labels.
Recovery shorter than online punish window. Low profiles cannot punish at 3–4 frame delay; players call the move “unfair safe.”
No meter cost. Spam at every neutral reset; approach becomes armor lottery.
Projectile height tags missing. Crush-high works inconsistently across characters using the same visual fireball.

Production checklist

Document power crush FSM phases in movelist CSV (vulnerable, crush, active, recovery).
Tag crush frames with hurtbox height override and one-hit armor flag.
Budget 2–4 vulnerable startup frames before armor opens.
Cap forward travel distance per move (1.5–2.5 body widths typical).
Assign meter cost (1–2 bars) or round cooldown.
Verify lows beat startup; throws beat armor; multi-hit breaks armor.
Match crush animation silhouette to hurtbox dip.
Tag projectiles with height band for crush interaction tests.
Recovery ≥ 6 frames launchable at target online delay.
Training mode overlay: crush frames, armor remaining, travel distance.
Telemetry: attempt rate, check success, low punish rate, conversion after hit.
Limit to 1–2 power crush tools per character for readability.

Key takeaways

Power crush combines advancing movement, one-hit armor, and crush-high hurtbox evasion — it is a neutral gap closer, not a startup-priority trade.
Vulnerable startup frames and long recovery are the counterplay layer; lows, throws, and launch punishes must remain viable.
Crush-high is hurtbox height geometry, distinct from crush counter and from stationary super armor.
Meter cost and travel caps prevent armor roulette and screen-crossing corner loops.
Harbor Brawl's refactor showed power crush can reopen midscreen without buffing rushdown damage numbers.