Azion V2 FANMADE

The AI behaves like an aggressive, predictive sword fighter. It constantly searches for the nearest living opponent and follows them as its main target. Instead of simply moving toward the opponent's current position, it watches how the opponent is moving and estimates their velocity and acceleration. It uses that information to predict where the opponent will be shortly, then moves and aims toward that predicted position. The AI tries to avoid moving in a completely straight line. It constantly changes its footwork using randomized behaviors such as strafing, circling, moving behind the opponent, moving to the side, moving diagonally, briefly hesitating, spinning, jumping, and suddenly changing direction. It has preferred attack-positioning behaviors. It often tries to get behind or beside the opponent instead of staying directly in front of them. Its preferred position is randomized, so its movement does not always look identical. The AI watches the opponent's movement and can react when the opponent suddenly slows down, changes direction, or turns. When this happens, it can temporarily change its movement pattern or reposition to another side. It also checks the surrounding space using directional sensing. This lets it determine which directions have more room and which directions are obstructed. It combines this information with its desired direction so it does not blindly move toward the opponent when that direction is unsafe or blocked. For aiming and facing, the AI does more than simply look directly at the opponent. It evaluates several nearby possible directions and recursively compares them for a short distance into the future. It prefers directions that keep the opponent in front of it, maintain useful positioning, have better clearance, and fit its current movement. Its turning is smoothed rather than instantly snapping toward the target. This makes its aim continuously adjust as the opponent moves. The AI also intentionally introduces small aim and movement variations. Its facing direction contains controlled random motion, so its aim is not perfectly identical every frame. Its movement contains momentum-like behavior. When it has been moving in a useful direction for a while, it can accumulate enough momentum in that direction to trigger a stronger side movement or repositioning maneuver. The AI periodically rerolls some of its movement and aiming parameters. This changes the exact timing, angle, jitter, and footwork pattern, making the behavior less repetitive. When the opponent gets close enough, the AI activates its weapon and attempts to attack. Overall, the behavior is: Observe opponent → track movement → predict future position → evaluate surrounding space → choose a combat direction → reposition around the opponent → adjust aim → attack when close → repeat continuously. The AI is not actually learning. It does not remember that a certain opponent beat one of its techniques and then permanently change its strategy. Its variety comes mainly from prediction, short-horizon directional evaluation, predefined behaviors, state changes, and weighted randomness. Its fighting style can therefore be described as: Aggressive + predictive + evasive + circling + randomized + short-term tactical.