Blender-Based 3D Cricket Bowling Animation with Pattern Recognition and Realistic Ball Trajectory Simulation

This paper presents an end-to-end virtual cricket system for bowling simulation and batting practice without any physical equipment. Using MediaPipe's pose estimation and Blender's physics engine, the system follows a bowler's movement through a standard webcam and generates real-time 3D animations from two viewpoints: a third-person view for bowlers to practice their skills and a first-person batsman's view for batting practice. The physics engine accurately simulates six fundamental delivery types-outswingers, inswingers, off-spin, leg-spin, yorkers, and bouncers-with spin/swing angle (±5°), pitch length (2-8m), and ball speed (60-160 km/h) parameters, replicating realistic ball motion through Magnus effect physics and bounce dynamics. Running at 30 FPS with 5s latency on consumer laptops, the system gives immediate feedback for both roles. Bowlers can precisely recreate different deliveries with 87% accuracy in recreating intended ball movement, while batsmen show 85% improvement in shot timing after repeated practice sessions. Markerless implementation prevents the requirement for specialized equipment, enabling cricket training using a single webcam. As the first end-to-end virtual cricket framework integrating bowling action capture with immersive batting practice, this work paves the way for new directions in equipment-free sports training. Future extensions will investigate multiplayer support and improved haptic feedback to further close virtual and physical training domains.