Guide

Game crush counter systems explained

Harbor Brawl's neutral patch shipped with generous counter-hit bonuses on every move, but slow heavy normals still lost to fast lights on simultaneous press. Players threw standing heavy punches to check dashes; opponents mashed jabs and won the exchange because both moves entered startup on the same frame and the faster active frame connected first. Replays labeled the outcome a “trade,” but the heavy user felt punished for spacing correctly. After tagging eligible heavies with an explicit crush counter flag that wins priority against opponent startup frames, grants a crumple state, and routes into wall splat near corners, intentional heavy checks rose 52% in ranked and neutral guess deaths from mashing fell 31%.

A crush counter is an attack property — not a universal combat state — that triggers when a qualifying move connects while the opponent is in move startup. The attacker's slow normal (or dedicated crush button) beats the defender's incoming button, applies a premium payoff (crumple, launch, extra damage, or forced knockdown), and often extends combo routes. Crush counters differ from passive counter-hits (bonuses when you hit any move during vulnerable frames) and from trade priority (simultaneous active-frame resolution). This guide covers eligibility rules, detection and priority, payoff taxonomy, corner routing, implementation, the Harbor Brawl refactor, a technique decision table, pitfalls, and a production checklist.

Crush counter vs counter-hit vs trade

Three systems overlap in players' minds but solve different design problems. Document them separately in your combat bible so balance patches do not break all three at once.

System What triggers it Who benefits Typical payoff
Counter-hit Any hit during defender startup or recovery Attacker (bonus on that hit) +damage, +hitstun, slide or launch on some moves
Crush counter Eligible move hits during defender startup only Crush-eligible attacker Crumple, wall splat, launch, hard knockdown
Trade / clash Both active frames overlap Priority rules or both take damage Symmetric damage or clash animation

Crush counters specifically reward reading commitment: the defender pressed a button; the attacker's slow check was already in motion and intercepts that startup. Counter-hits reward timing more broadly (including recovery). Trades resolve when neither side had crush or invincibility priority. A single exchange can be labeled crush counter + counter-hit in games that stack flags, but for clarity Harbor Brawl applies crush properties only on the opening hit and routes follow-ups under normal scaling.

Move eligibility and risk budget

Not every move should crush. Eligibility is how you price risk in neutral: slow, committal attacks earn startup priority; fast lights do not.

Common eligibility rules

  • Startup threshold — only moves with 12+ startup frames (tune per game speed) can crush. Lights at 4–6f stay honest.
  • Move class whitelist — standing/crouching heavies, designated “crush” buttons, or slow command normals. Exempt throws and invincible reversals.
  • Whiff risk — crush-eligible moves should have visible recovery if they whiff; the intercept reward pays for that whiff vulnerability.
  • Projectile exclusion — most games exclude fireballs from crushing physical startup; otherwise zoning dominates reads.

Pair eligibility with frame data published for tournament players. If a heavy crushes at 15f startup, document which jabs (4–5f) and specials (8–12f) it beats on reaction vs on anticipation.

Detection pipeline at impact

On hitbox–hurtbox overlap, evaluate in deterministic order for rollback safety:

  1. Defender state — is MOVE_STARTUP true? Recovery and idle block do not qualify for crush (only counter-hit if applicable).
  2. Attacker move flag — does the move carry CAN_CRUSH_COUNTER?
  3. Priority resolution — if both are in startup, crush-eligible attacker wins over non-eligible defender startup; two crush moves may use speed tie-break or mutual clash.
  4. Payoff selection — apply crumple, launch, or wall route based on move tier, spacing, and wall proximity.

Store crush as a hit property on the opening frame, not as a persistent buff. Follow-up hits in the same combo use normal hitstun unless you explicitly design crush-only extensions (rare; often overpowered).

Payoff taxonomy

Crush payoffs must be readable and route into your existing knockdown and wall systems without bespoke code per character.

Crumple

Defender folds to knees with extended hitstun, standing hurtbox, no hard knockdown. Enables high/low mixups, delayed throws, or safe jump setups. Crumple duration scales with crush move tier (heavy vs EX crush). Always pair with a visible floor impact effect — players learn crush from the crumple animation faster than from HUD text.

Launch and float

Airborne state for juggle extensions. Use when crush move is a launcher or when corner height allows juggle routes without wall splat. Cap float time so crush into launch does not exceed your combo damage budget.

Wall splat and carry

Near stage edge, crush can force wall splat instead of crumple mid-screen. Integrate with wall HP or splat decay so repeated crush loops into the same corner do not become infinite. Mid-screen crush into crumple, corner crush into splat is a proven readability pattern.

Hard knockdown and slide

Lower-ceiling payoffs for crush moves that are already plus on block. Slide knockdowns extend oki without full crumple mixup density — good for characters with weak overhead game.

Neutral and character design implications

Crush counters reshape footsies without adding a new button if heavies carry the property. Characters with excellent 4f lights become crushable on mash; characters with 15f heavies become honest checks against dash and jump startup.

  • Buffer interactions — define whether crush beats buffered specials that have not reached active frames. Most games crush physical startup only, not full input buffer windows.
  • Armor and super armor — crush priority typically loses to armor on active frames; crushing armor startup before active is a common skill test.
  • Jump startup — crushing jump startup (pre-landing invuln) rewards anti-air reads; crushing empty jump can feel oppressive — gate with range or require counter-hit height.
  • Meter and EX crush — optional EX layer: spend meter for crush property on a medium normal. Raises skill ceiling but needs audible telegraph so it is not a neutral lottery.

Harbor Brawl refactor

Before crush flags, Harbor Brawl resolved same-frame startup with fastest active frame wins — heavies never traded favorably. Neutral devolved into light wars. The refactor:

  • Tagged standing/crouching heavies and two slow command normals with CAN_CRUSH_COUNTER; lights and mediums excluded.
  • On crush trigger: mid-screen crumple 28f; within 120px of wall, wall splat with 22f stick instead.
  • Distinct gold spark VFX + unique SFX layer on crush (separate from blue counter-hit flash).
  • Training mode: crushable window shaded on opponent move startup in frame overlay.
  • Disabled crush on simultaneous crush vs crush — mutual clash reset to neutral spacing.

Heavy check success rate in neutral rose from 38% to 58% against mash-heavy opponents. Average round length increased 8% because crush routes converted into real damage instead of losing to 4f trades. Complaints about “random losses” on heavy press dropped sharply once crush feedback shipped.

Technique decision table

Design goal Prefer crush counter system Prefer alternative
Reward slow normals in neutral Crush on heavies vs startup Flat damage buff on heavies only
Punish mashing without invincible reversals Crush intercept on 4f lights Frame 1 invincible DP (higher skill floor variance)
Corner routing and spectacle Crush into wall splat near edge Corner-only command grab
Simple casual fighter Crush on one universal heavy per character Counter-hit damage only (no crumple)
Prevent neutral stalemates Crush + visible crumple mixup Universal dash cancel on all normals

Common pitfalls

  • Crush on fast moves. If 6f mediums crush, lights become useless and neutral homogenizes around medium speed.
  • Crumple without escape. Infinite stand overhead vs crouch jab loops after crush feel unfair; add delayed stand-up invuln or pushback.
  • Stacking crush + full counter-hit bonuses. Double-stacking damage and stun blows combo budgets; pick one primary reward.
  • Invisible crush windows. Same failure mode as counter-hits — always ship training overlays and distinct VFX.
  • Crush beating projectiles. Accidentally crushing fireball startup from across screen breaks zoning roles.
  • Rollback ambiguity. Crush must resolve from confirmed startup flags on both sides, not animation pose guesses.
  • Wall splat every crush. Removes mid-screen reward variety; use proximity gates.

Production checklist

  • Define crush eligibility: startup threshold, move whitelist, projectile rules.
  • Document detection order: defender startup, attacker flag, priority tie-breaks.
  • Specify payoffs per tier: crumple duration, launch height, wall splat range.
  • Cap crush damage contribution under combo proration and stun scaling caps.
  • Integrate wall splat decay and wall-break meters for corner loops.
  • Separate VFX/SFX from counter-hit and trade feedback.
  • Expose crush windows in training mode and replay scrubbers.
  • Test buffer, same-frame crush vs light, and armor interaction edge cases.
  • Validate crush flags in rollback simulation across 100+ random exchanges.
  • Publish crush-eligible move list alongside frame data for each character.
  • Balance patch note template: which heavies gained or lost crush property.
  • Telemetry: crush rate per move, conversion to damage, wall route frequency.

Key takeaways

  • Crush counters are move properties that beat opponent startup — distinct from universal counter-hit bonuses and trade priority.
  • Eligibility belongs on slow, committal normals so fast lights keep honest risk/reward.
  • Crumple, wall splat, and launch payoffs should plug into existing knockdown and wall systems, not one-off scripts.
  • Harbor Brawl's crush flags raised heavy check success 38% to 58% and cut mash deaths 31% with gold-spark feedback.
  • Readable training overlays matter as much as frame math — players cannot learn crush without seeing startup windows.

Related reading