GENERATIONS · BUILD 45 / GAMEPLAY 196
Combat data you can inspect, question and build on.
Search the installed build’s authored parameters, trace damage references, and download a queryable database. Every raw value retains its source. Effective damage and frame timings remain unverified.
What this can establish today: which fields and skill references the shipped assets contain. Radius, velocity and script-frame arguments use unvalidated engine units. What it cannot establish yet: strongest character, guaranteed punish, true combo, per-hit damage or tier placement. The three-attack health pilot remains a provisional observation.
Explore the data
How to use this for competitive research
- Find the action. Use the move inventory for localized inputs and action slots for literal authored references. A slot name does not prove runtime dispatch.
- Trace the hypothesis. Find a skill’s
Hit.damageId, then search that exact identifier in Damage identifiers. Raw bytes are retained; no float offset is labeled damage without validation. - Test one condition. Record the exact build, effective settings, fighters/forms, guard state, distance, resources and input sequence. Preserve successful and failed attempts.
- Measure a decision. For a punish, measure both players’ actionable updates. For a combo, test escape windows and net resource cost. For neutral, test range, tracking and whiff recovery.
- Compare like conditions. Keep forms, balance settings, matchups and stages separate. Tier lists also need matchup evidence and player results; move damage alone is insufficient.
Open the existing Player Lab and choose Practice log to record community observations. Entries remain user-reported until independently checked.
Scope and identity issues
Blank and END slot records remain in the full database. This browser shows named non-END slots. The inventory contains internal groups; it is not a reconciled playable roster.