Blending Additive Hit Reactions with Locomotion

In modern game development, maintaining character responsiveness while delivering visceral combat feedback requires blending hit-react animations directly on top of active walking locomotion cycles. This technique relies on additive animation blending, bone-specific masking, and dynamic physics stabilization to ensure that a character flinches realistically without interrupting their movement direction or causing foot sliding. This article explores the technical pipeline animators and technical designers use to calculate additive deltas, isolate skeletal hierarchies, dynamically map hit directions, and smoothly return characters to their base locomotion states.

Understanding Additive Animation

Standard animations rely on absolute transform values for skeletal joints in local or component space. An additive animation, however, stores the difference (delta) between an animated pose and a static reference pose.

To create an additive hit-reaction:

  1. The animator authors a full hit-reaction flinch starting from a standardized base pose (commonly a neutral T-pose, A-pose, or a standard idle).
  2. The game engine subtracts the reference pose from the hit-reaction frame-by-frame: Additive Pose = Hit React Pose - Reference Pose.
  3. During runtime, the engine adds these delta transforms onto the current frame of the character's active walking locomotion cycle: Final Pose = Locomotion Pose + (Additive Pose * Weight).

Because only the delta changes in rotation, translation, and scale are calculated, the hit reaction applies as an offset without overriding the cyclic motion of the character's skeleton.

Layered Skeletal Masking

Applying an additive hit reaction to the entire skeleton during locomotion often leads to visual artifacts, primarily foot sliding and broken leg kinematics. To prevent this, animators use layered bone masking (often referred to as a "Layered blend per bone" node in Unreal Engine or an "Avatar Mask" in Unity).

The skeleton is divided into functional zones:

  • The Upper Body (Spine, Chest, Head, Shoulders, Arms): Receives 100% of the additive hit-react weight. The upper torso absorbs the impact, twists back, or drops based on the strike.
  • The Lower Body (Pelvis, Legs, Feet): Retains 0% to 10% of the additive weight. The legs continue their walk cycles driven entirely by the base locomotion blend space.
  • The Pelvis Transition: The pelvis acts as the root of the skeletal hierarchy. Animators typically blend the pelvis at a reduced weight (e.g., 20–30%) or damp its rotational delta. This allows the hips to react slightly to a heavy blow without pulling the feet off their grounded stride paths.

Directional Hit Mapping

Seamless hit reactions require directional context. If an enemy strikes a walking player from behind, flinching backward breaks immersion.

Animators author hit reacts in multiple cardinal directions—typically Front, Back, Left, and Right. At runtime:

  1. The game engine calculates the dot product and cross product between the character's forward vector and the incoming damage vector.
  2. This calculation yields an angle (from -180 to 180 degrees) relative to the character’s facing direction.
  3. A 2D blend space dynamically interpolates between the directional additive animations based on that angle.

Because the system is additive, whether the character is walking forward, backward, or strafing, the directional hit react aligns naturally with the impact source relative to the character's current torso rotation.

Transition Curves and Inertialization

How the animation starts and finishes determines whether the blend feels seamless or robotic. If an additive hit react snaps in too quickly, it looks like a visual glitch; if it blends too slowly, the impact loses its punch.

  • Attack and Decay Curves: Hit reactions typically use a very sharp attack (blending in over 0.05 to 0.1 seconds) to simulate the sudden transfer of kinetic energy, followed by a gradual decay (blending out over 0.3 to 0.6 seconds) as the character recovers.
  • Inertial Blending (Inertialization): Traditional crossfades blend two active poses over time, which can create a "mushy" double-exposure look during fast movement. Modern engines use inertialization, which captures the velocity and momentum of the skeleton at the moment the hit react ends and computes an inertial decay toward the current locomotion state. This guarantees smooth recovery without foot-skating or abrupt skeletal snapping.

Inverse Kinematics Integration

Even with careful masking, upper-body flinches shift the character’s center of mass and spine angle. To ensure the feet remain planted firmly on uneven terrain during a hit reaction, foot Inverse Kinematics (IK) passes must execute after the additive animation is evaluated.

By applying foot IK, two-bone IK, and pelvis offset correction at the end of the post-process animation graph, the character’s feet lock solidly to the ground surface, preserving clean footstep contacts while the upper body twists and recovers from the strike.