Genichiro · combat reconstruction

The project brings player actions, boss decisions, contact rules and presentation into a shared UE runtime. The film shows a complete three-phase Genichiro encounter, including counters, phase transitions and the final deathblow. The presentation focuses on the actions and systems of this Boss encounter.

Armored Genichiro and the Way of Tomoe phase use separate decision policies. Distance, relative direction, posture, status effects, timers and available space change the weights of possible actions. Close exchanges are evaluated before ordinary action selection, allowing a defensive response to lead into the next attack.

Selected actions run through goal steps for approach, attacks, repeated combos, retreat and lateral movement. Conditions, callbacks and completion return control to the decision loop. This connects the choice of a move to its execution and recovery across the encounter.

Action data and timed events define attack windows and defensive opportunities. Contact feeds health and posture changes, guard or deflection reactions, movement, sound and effects. Events belong to a particular action and contact lifetime, so an interrupted move can finish its own feedback without overwriting a new one.

Input capture and action execution use distinct windows. Parallel reaction layers and movement each retain their own state, keeping a visible animation from becoming the sole authority for combat rules.

Mikiri counters and paired deathblows coordinate both actors’ actions, position and direction while passing collision results back into the shared motion. Full posture is one condition; preparation, contact and execution also have their own permissions.

Lightning reversal tracks reception, accepted action, return events and effect release as parts of the same interaction. The showcase presents these systems in the three-phase battle, alongside the actions and recovery that connect them.

Combat state, paired animations and the transition sequence have distinct lifetimes. Completing an interaction and completing the body animation are handled separately before the next phase takes ownership. Camera targets are updated after both actors’ poses, keeping the viewpoint aligned with the current encounter.